1
00:00:00,000 --> 00:00:04,210
Audrow Nash: I'm sure you've
heard about the Robot Operating

2
00:00:04,210 --> 00:00:08,280
System, or ROS. If you haven't,
it's freely available open

3
00:00:08,280 --> 00:00:12,349
source software that makes it
much easier to build robots.

4
00:00:12,349 --> 00:00:16,489
What you may not know is that
the Robot Operating System is

5
00:00:16,489 --> 00:00:20,910
one of the projects maintained
by Open Robotics. Other projects

6
00:00:20,910 --> 00:00:25,471
include the Gazebo simulator and
OpenRMF. Open Robotics had a non

7
00:00:25,471 --> 00:00:29,540
profit part and a for profit
part. At the end of 2022, the

8
00:00:29,540 --> 00:00:34,031
for profit part of Open Robotics
was bought by Intrinsic AI, and

9
00:00:34,031 --> 00:00:37,750
we've been maintaining and
investing in Open Robotics

10
00:00:37,750 --> 00:00:42,030
projects from Intrinsic, which
you'll hear about more in this

11
00:00:42,030 --> 00:00:45,889
interview. I say we because I
was at Open Robotics as a

12
00:00:45,889 --> 00:00:49,748
software engineer and I came
over to Intrinsic with the

13
00:00:49,748 --> 00:00:54,099
acquisition. Anyways, it's been
a little over a year since the

14
00:00:54,099 --> 00:00:58,239
acquisition and now the non
profit part of Open Robotics is

15
00:00:58,239 --> 00:01:02,449
announcing a new governance
structure called the Open Source

16
00:01:02,449 --> 00:01:06,589
Robotics Alliance, or OSRA,
which is the main focus of this

17
00:01:06,589 --> 00:01:10,869
interview. From my perspective,
this is a great move for Open

18
00:01:10,869 --> 00:01:15,149
Robotics' projects and follow
successful open source software

19
00:01:15,149 --> 00:01:19,710
projects like those of the Linux
Foundation. I also think it will

20
00:01:19,710 --> 00:01:23,639
be better for companies that
rely heavily on any of Open

21
00:01:23,639 --> 00:01:27,498
Robotics' projects. And I
believe it will be better for

22
00:01:27,498 --> 00:01:31,919
contributors to the community as
OSRA formalizes the mentorship

23
00:01:31,919 --> 00:01:35,568
process and has a way for
contributors to have their

24
00:01:35,568 --> 00:01:39,988
voices heard. Also, a shout out
to OSRA's founding sponsors. At

25
00:01:39,988 --> 00:01:44,198
the Platinum level, we have
Intrinsic, NVIDIA, and Qualcomm;

26
00:01:44,198 --> 00:01:48,338
at Gold, Apex, and Zetascale; at
Silver, Clearpath, Ekumen,

27
00:01:48,338 --> 00:01:52,548
eProsima, and Picnic. And then
Silicon Valley Robotics is an

28
00:01:52,548 --> 00:01:56,688
associate, and Canonical, and
Open Navigation as supporting

29
00:01:56,688 --> 00:02:00,828
organizations. I think you'll
like this interview if you're

30
00:02:00,828 --> 00:02:05,248
interested in the Open Robotics
and Willow Garage lore, want to

31
00:02:05,248 --> 00:02:08,967
learn more about the Open
Robotics projects after the

32
00:02:08,967 --> 00:02:13,317
Intrinsic Acquisition, or want
to be part of the next stage of

33
00:02:13,317 --> 00:02:17,317
development for the Robot
Operating System or any of Open

34
00:02:17,317 --> 00:02:21,527
Robotics' other projects. I've
included a link to learn more

35
00:02:21,527 --> 00:02:25,877
about OSRA in the description,
and with that, I hope you enjoy

36
00:02:25,877 --> 00:02:27,000
the interview. ​

37
00:02:27,000 --> 00:02:31,000
Audrow Nash: Tully, would you
introduce yourself?

38
00:02:31,000 --> 00:02:35,186
Tully Foote: Hi, I'm Tully
Foote, the community advisor to

39
00:02:35,186 --> 00:02:38,000
OSRF and open source lead at
Intrinsic.

40
00:02:38,000 --> 00:02:40,000
Audrow Nash: Awesome. And Geoff.

41
00:02:40,000 --> 00:02:43,441
Geoff Biggs: Hi, I'm Geoff
Biggs. I am the CTO of the Open

42
00:02:43,441 --> 00:02:46,466
Source Robotics Foundation,
otherwise known as Open

43
00:02:46,466 --> 00:02:47,000
Robotics.

44
00:02:47,000 --> 00:02:51,455
Audrow Nash: Awesome, and so I
would like to go. Back to the

45
00:02:51,455 --> 00:02:55,985
beginning and talk about Open
Robotics beginning and starting

46
00:02:55,985 --> 00:03:00,515
with Willow Garage. So Tully,
would you start with telling me

47
00:03:00,515 --> 00:03:02,000
about Willow Garage?

48
00:03:02,000 --> 00:03:06,157
Tully Foote: Yeah, Willow Garage
was a research, a robotics

49
00:03:06,157 --> 00:03:10,597
research company in Menlo Park.
I joined it right out of school

50
00:03:10,597 --> 00:03:14,402
and it was really a great
experience, a lot of what we

51
00:03:14,402 --> 00:03:18,630
were driven for was mission and
impact. And through that, we

52
00:03:18,630 --> 00:03:22,646
started the ROS project, or at
Willow, we started the ROS

53
00:03:22,646 --> 00:03:26,592
project in collaboration with
Stanford and several other

54
00:03:26,592 --> 00:03:30,750
partners and universities. And
it was a great time. We were

55
00:03:30,750 --> 00:03:34,907
able to go out and build the
infrastructure that we thought

56
00:03:34,907 --> 00:03:38,994
was the most important to be
able to bring robotics to the

57
00:03:38,994 --> 00:03:42,729
greater community, in
particular, leveraging the open

58
00:03:42,729 --> 00:03:46,322
source ethos. To help with
sharing and communicate,

59
00:03:46,322 --> 00:03:49,000
building a community around the
world.

60
00:03:49,000 --> 00:03:53,581
Audrow Nash: Hell yeah. And what
was it just to make it a bit

61
00:03:53,581 --> 00:03:58,162
more tangible? It was centered
around a PR2 robot. That large

62
00:03:58,162 --> 00:04:02,668
robot with arms and the kind of
really iconic, I don't know,

63
00:04:02,668 --> 00:04:06,648
depth camera on the top or
stereo. So it was centered

64
00:04:06,648 --> 00:04:08,000
around the PR2 and

65
00:04:08,000 --> 00:04:11,944
Tully Foote: Yeah. So the PR2
was our main platform. We were

66
00:04:11,944 --> 00:04:15,625
really focused on having impact
in the robotics research

67
00:04:15,625 --> 00:04:19,832
community. And we wanted to have
a common platform for research,

68
00:04:19,832 --> 00:04:23,841
researchers around the world in
robotics. And so we built the

69
00:04:23,841 --> 00:04:27,851
PR2 and were able to distribute
it around the world. And with

70
00:04:27,851 --> 00:04:31,992
the PR2, we also built ROS to be
the software that people could

71
00:04:31,992 --> 00:04:36,067
collaborate with. Because one of
the things is, if we give you

72
00:04:36,067 --> 00:04:39,551
both robots and you run
different software, it's very

73
00:04:39,551 --> 00:04:43,035
hard to collaborate. but by
providing an open source,

74
00:04:43,035 --> 00:04:46,979
reference design for the PR2
that everyone could pick up and

75
00:04:46,979 --> 00:04:51,054
build, we actually had people
that received PR2s, they've been

76
00:04:51,054 --> 00:04:55,064
practicing in simulation using
the open source software. They

77
00:04:55,064 --> 00:04:58,548
extended it with their research
and were able to have

78
00:04:58,548 --> 00:05:02,689
publishable research. The week
they received their robot, which

79
00:05:02,689 --> 00:05:06,567
was very different than the
state of the art prior to that.

80
00:05:06,567 --> 00:05:09,722
Or each major research
institution had their own

81
00:05:09,722 --> 00:05:13,600
bespoke framework for doing
their integration tasks. And if

82
00:05:13,600 --> 00:05:17,676
you wanted to reuse or try to
reproduce the results of another

83
00:05:17,676 --> 00:05:21,357
university, you generally had to
re implement their full

84
00:05:21,357 --> 00:05:23,000
algorithm and try it out.

85
00:05:23,000 --> 00:05:28,207
Audrow Nash: And what timeframe
was this? Like about when was

86
00:05:28,207 --> 00:05:30,000
Willow Garage active?

87
00:05:30,000 --> 00:05:36,725
Tully Foote: Willow Garage was
active from late 2007 until

88
00:05:36,725 --> 00:05:38,000
about 2013.

89
00:05:38,000 --> 00:05:39,000
Audrow Nash: Awesome.

90
00:05:38,000 --> 00:05:41,803
Audrow Nash: And Tully, you've
mentioned ROS, which is the

91
00:05:41,803 --> 00:05:45,541
robot operating system, but
Geoff, you were involved with

92
00:05:45,541 --> 00:05:49,541
Player at the very beginning, so
tell me. What Player was and

93
00:05:49,541 --> 00:05:53,148
then how that turned into ROS
and how you came in to be

94
00:05:53,148 --> 00:05:54,000
involved too.

95
00:05:54,000 --> 00:05:57,560
Geoff Biggs: I will try to be
brief. It's a long history. if

96
00:05:57,560 --> 00:06:01,120
you go right back to 1999, back
then every robot had its own

97
00:06:01,120 --> 00:06:04,858
software from the maker and they
had very fixed ideas about how

98
00:06:04,858 --> 00:06:08,595
their robot should be used. and
It was pretty difficult, people

99
00:06:08,595 --> 00:06:12,333
trying to share code and so on.
and then a bunch of guys at USC

100
00:06:12,333 --> 00:06:15,834
Southern California, I think it
was, one of them was named,

101
00:06:15,834 --> 00:06:19,512
Brian Gerkey, who you may have
heard of. he, they came up with

102
00:06:19,512 --> 00:06:23,191
this idea for a simulator, and
then on top of that, they built

103
00:06:23,191 --> 00:06:26,869
this thing called Player, which
allowed you to basically write

104
00:06:26,869 --> 00:06:30,429
software for your robot. And
share it with other people, and

105
00:06:30,429 --> 00:06:33,811
they thought, hey, this is
pretty cool, let's open source

106
00:06:33,811 --> 00:06:37,311
it and see what happens. And
they did, and it exploded. the

107
00:06:37,311 --> 00:06:40,990
whole world started using it.
There was a point in about 2003,

108
00:06:40,990 --> 00:06:44,194
2004, if you went to a
conference and said you weren't

109
00:06:44,194 --> 00:06:47,932
using Player, people would look
at you like you were stupid. it

110
00:06:47,932 --> 00:06:51,610
was, like ROS is now, back then,
and this is 20 years ago now,

111
00:06:51,610 --> 00:06:55,229
it's amazing to think about, I
myself, during my PhD, back in

112
00:06:55,229 --> 00:06:58,908
about 2003 and 4, and one of my
lab mates, we were looking for

113
00:06:58,908 --> 00:07:02,290
new software for our robot,
because the existing software

114
00:07:02,290 --> 00:07:05,849
was pretty awful, and we came
across Player, and we tried it

115
00:07:05,849 --> 00:07:09,231
out, and it was pretty cool, but
it had some problems, it

116
00:07:09,231 --> 00:07:12,494
couldn't handle shifting sensor
frames, which was a bit

117
00:07:12,494 --> 00:07:16,232
annoying. And so we decided, oh,
we'll fix that. And as part of

118
00:07:16,232 --> 00:07:19,673
fixing it, we, particularly my
lab mate, a guy called Toby

119
00:07:19,673 --> 00:07:23,352
Collett, he pretty much rewrote
the entirety of Player. And we

120
00:07:23,352 --> 00:07:26,793
sent it into Brian and said,
hey, we've made it better and

121
00:07:26,793 --> 00:07:30,234
fixed all these problems. You
guys are awesome. You can be

122
00:07:30,234 --> 00:07:33,913
developers now. So we ended up
being developers on Player. And

123
00:07:33,913 --> 00:07:37,473
so we ended up contributing to
Player for a few years. Which

124
00:07:37,473 --> 00:07:41,210
was pretty cool. And that's how
I got to know Brian. And, a few

125
00:07:41,210 --> 00:07:44,830
years later, Brian sent me an
email saying, Hey, I'm starting

126
00:07:44,830 --> 00:07:48,389
this new company, Willow Garage,
and we're making this thing

127
00:07:48,389 --> 00:07:51,000
called ROS. It's going to be
pretty awesome.

128
00:07:51,000 --> 00:07:55,260
Audrow Nash: Hell yeah. And
just, how would you describe, so

129
00:07:55,260 --> 00:07:59,734
for those who are not familiar
with ROS, Geoff, what was Player

130
00:07:59,734 --> 00:08:03,000
and like what task did it
accomplish that made

131
00:08:03,000 --> 00:08:06,596
Geoff Biggs: It was Player was
Yeah, Player was It was like an

132
00:08:06,596 --> 00:08:09,787
early Iteration of ROS in some
respects. So ROS is very

133
00:08:09,787 --> 00:08:13,209
distributed. Every single, node
is its own process usually,

134
00:08:13,209 --> 00:08:16,747
right? When ROS is yours, you
can compose it now. Player, all

135
00:08:16,747 --> 00:08:19,763
of those are what we call
drivers, and they were all

136
00:08:19,763 --> 00:08:23,244
loaded into the Player server.
So they're all built into one

137
00:08:23,244 --> 00:08:26,086
binary and you had a
configuration file that said

138
00:08:26,086 --> 00:08:29,740
what drivers to load and so on
and they talked by drivers would

139
00:08:29,740 --> 00:08:33,162
publish information and they
could subscribe to information

140
00:08:33,162 --> 00:08:36,295
so it had a predecessor to
topics. I mean it's similar

141
00:08:36,295 --> 00:08:39,891
concepts in terms of publish and
describe but different way of

142
00:08:39,891 --> 00:08:43,371
achieving it and so if you had
these you have player clients

143
00:08:43,371 --> 00:08:47,142
which you set on your desktop or
on another robot or whatever and

144
00:08:47,142 --> 00:08:50,390
player servers which ran on the
robot to control all the

145
00:08:50,390 --> 00:08:53,000
hardware based on the commands
they received.

146
00:08:53,000 --> 00:08:57,473
Audrow Nash: Gotcha. Yeah, it's
so cool to see how we're making,

147
00:08:57,473 --> 00:09:01,737
the evolution of software to
become more and more reusable so

148
00:09:01,737 --> 00:09:05,441
people don't have to keep
reinventing the wheel. Just

149
00:09:05,441 --> 00:09:06,000
awesome.

150
00:09:06,000 --> 00:09:11,727
Audrow Nash: And then, back to
Tully. Okay, so Willow Garage is

151
00:09:11,727 --> 00:09:17,000
operating. How does Willow
Garage turn into Open Robotics?

152
00:09:17,000 --> 00:09:21,059
Tully Foote: Yeah, so we, at
Willow Garage, we spent a lot of

153
00:09:21,059 --> 00:09:25,117
time, we were able to, build ROS
up, build the community, but

154
00:09:25,117 --> 00:09:29,109
there were a lot of people that
were rightfully a little bit

155
00:09:29,109 --> 00:09:33,035
worried about, what Willow
Garage's intentions were. it's a

156
00:09:33,035 --> 00:09:37,226
for profit company and we were
building out, we were doing open

157
00:09:37,226 --> 00:09:41,285
source work and proving a very
good track record and building

158
00:09:41,285 --> 00:09:45,077
community and getting people
working with us. but what we

159
00:09:45,077 --> 00:09:49,203
wanted to do was make sure that
there was a, community steward

160
00:09:49,203 --> 00:09:53,394
that could be a little bit more
independent than Willow Garage.

161
00:09:53,394 --> 00:09:57,652
Willow Garage was a building the
PR2, looking at other products.

162
00:09:57,652 --> 00:10:01,179
there were seven different
companies that spun out of

163
00:10:01,179 --> 00:10:05,370
Willow Garage. And of those, one
of the ones we created was the

164
00:10:05,370 --> 00:10:09,496
Open Source Robotics Foundation.
And this is, was set up to be

165
00:10:09,496 --> 00:10:13,355
the hub and steward of the
community as a non profit, such

166
00:10:13,355 --> 00:10:17,347
that everyone know that this is
going to be a neutral party.

167
00:10:17,347 --> 00:10:21,072
That's there, and specifically
focused on the mission of

168
00:10:21,072 --> 00:10:24,000
promoting open source software
and robotics.

169
00:10:24,000 --> 00:10:28,607
Audrow Nash: And why is it
important to have a steward?

170
00:10:28,607 --> 00:10:33,634
what's, why, was this a risk to
companies who might consider

171
00:10:33,634 --> 00:10:38,995
using it? Or like, why, does the
community need a steward? What,

172
00:10:38,995 --> 00:10:40,000
is the role?

173
00:10:40,000 --> 00:10:43,790
Tully Foote: The steward is not
necessary. There are a lot of

174
00:10:43,790 --> 00:10:47,643
good open source projects that
are maintained and operated by,

175
00:10:47,643 --> 00:10:51,434
for profit entities. But the
neutral steward helps facilitate

176
00:10:51,434 --> 00:10:55,348
the community building. One of
the things you could potentially

177
00:10:55,348 --> 00:10:59,014
worry about would be if, the
project is controlled by a for

178
00:10:59,014 --> 00:11:02,929
profit entity, they can drive it
directly in the direction that

179
00:11:02,929 --> 00:11:06,720
they want. For their purposes,
which might be, competing with

180
00:11:06,720 --> 00:11:10,386
you. So if you consider you're
contributing to this project

181
00:11:10,386 --> 00:11:13,803
that somebody, some other
company that is potentially a

182
00:11:13,803 --> 00:11:17,656
direct competitor controls, you
might be worried about whether

183
00:11:17,656 --> 00:11:21,322
or not your support of the
project is actually helping them

184
00:11:21,322 --> 00:11:23,000
more than it's helping you.

185
00:11:23,000 --> 00:11:24,000
Audrow Nash: Okay.

186
00:11:23,000 --> 00:11:24,000
Geoff Biggs: If I can

187
00:11:24,000 --> 00:11:26,000
Audrow Nash: would also think,
oh yeah, go ahead.

188
00:11:26,000 --> 00:11:29,288
Geoff Biggs: there is also,
something I've learned in the

189
00:11:29,288 --> 00:11:33,038
past year. I never thought about
it much before that, but there's

190
00:11:33,038 --> 00:11:36,096
also the, the legal IP
protection as well is actually

191
00:11:36,096 --> 00:11:39,096
surprisingly important for
companies, like logos and

192
00:11:39,096 --> 00:11:42,788
trademarks and all that sort of
stuff. Companies care about that

193
00:11:42,788 --> 00:11:46,077
apparently, so we have to
protect that and it's better to

194
00:11:46,077 --> 00:11:49,308
have that done by a neutral
steward. Then by one company

195
00:11:49,308 --> 00:11:52,712
they might suddenly design
licensing fees or something like

196
00:11:52,712 --> 00:11:53,000
that.

197
00:11:53,000 --> 00:11:58,598
Audrow Nash: Oh, okay. So
there's all that. And then, the

198
00:11:58,598 --> 00:12:04,786
perspective of being maintainers
and developers of the code, so

199
00:12:04,786 --> 00:12:11,170
OSRF, now that was the nonprofit
arm. When did OSRC come to be? I

200
00:12:11,170 --> 00:12:15,000
don't, know, maybe Tully, do you
think?

201
00:12:15,000 --> 00:12:16,000
Tully Foote: Yeah.

202
00:12:16,000 --> 00:12:18,000
Audrow Nash: profit side of Open
Robotics.

203
00:12:18,000 --> 00:12:22,243
Tully Foote: the way that OSRF
operated was that it did

204
00:12:22,243 --> 00:12:26,485
consulting services for
companies to pay for the staff.

205
00:12:26,485 --> 00:12:31,191
And in the overhead, we also did
the open source maintenance.

206
00:12:31,191 --> 00:12:35,588
This is a familiar model for
many open source projects. I

207
00:12:35,588 --> 00:12:40,447
think that the best analogy for
this is Mozilla Corporation did

208
00:12:40,447 --> 00:12:45,230
a very similar thing. It started
off as a non profit. but then

209
00:12:45,230 --> 00:12:49,935
your non profit status can be
threatened by, if you're taking

210
00:12:49,935 --> 00:12:54,795
too much contracting consulting
work, which could be considered

211
00:12:54,795 --> 00:12:59,424
work for not the benefit of the
public, but the benefit of a

212
00:12:59,424 --> 00:13:04,206
specific company or individual.
And so the consulting arm was,

213
00:13:04,206 --> 00:13:08,295
spun out into a separate
company. Where the contracts

214
00:13:08,295 --> 00:13:13,000
went through and that's where
all the staffers were employed.

215
00:13:13,000 --> 00:13:16,493
Audrow Nash: Awesome. Okay. So
that's very clear. So what were

216
00:13:16,493 --> 00:13:17,000
you guys?

217
00:13:17,000 --> 00:13:21,474
Audrow Nash: So you, Tully, you
joined right out of school.

218
00:13:21,474 --> 00:13:26,251
Geoff, you came along at Brian's
invitation. What were you guys

219
00:13:26,251 --> 00:13:30,725
involved with at Open Robotics?
What was your trajectory of

220
00:13:30,725 --> 00:13:33,000
roles? Maybe start with Geoff.

221
00:13:33,000 --> 00:13:36,819
Geoff Biggs: Mine's actually a
bit different from Tully's, so

222
00:13:36,819 --> 00:13:40,576
I, I didn't join Willow Garage
at all, and I didn't actually

223
00:13:40,576 --> 00:13:44,457
join Open Robotics until 2020, I
think. It was halfway through

224
00:13:44,457 --> 00:13:48,026
the pandemic. Yeah. so Brian
invited me right back at the

225
00:13:48,026 --> 00:13:51,970
beginning, but at that point I
was doing a postdoc in Japan and

226
00:13:51,970 --> 00:13:55,351
quite enjoying myself. I felt
like I want to really, I

227
00:13:55,351 --> 00:13:59,045
literally just started the
postdoc three months earlier and

228
00:13:59,045 --> 00:14:03,052
I felt I want to give this a go
for a couple of years. and then,

229
00:14:03,052 --> 00:14:06,809
the offer was, Brian said the
offer's always open. and I was

230
00:14:06,809 --> 00:14:10,252
working as a researcher in
Japan. I thought I was quite

231
00:14:10,252 --> 00:14:13,508
happy, and then, I found a
reason not to leave Japan

232
00:14:13,508 --> 00:14:17,390
because, I got a girlfriend and
then got married. so that kind

233
00:14:17,390 --> 00:14:21,083
of kept me glued here for a
while, and then after a while I

234
00:14:21,083 --> 00:14:24,652
got sick of how the research
world of robotics was going,

235
00:14:24,652 --> 00:14:28,221
just a bit too much pointless
papers, I felt, rather than

236
00:14:28,221 --> 00:14:32,290
doing actual real world impacts,
and I wanted to be more involved

237
00:14:32,290 --> 00:14:36,110
in what I saw is like basically
the robot revolution that was

238
00:14:36,110 --> 00:14:39,741
starting to happen mostly thanks
to ROS, you know That was

239
00:14:39,741 --> 00:14:43,184
something I want to be more
involved in and so I got an

240
00:14:43,184 --> 00:14:47,066
offer from a startup in Japan
And I went to join that And then

241
00:14:47,066 --> 00:14:50,948
after about a year and a half of
that, I figured that you know

242
00:14:50,948 --> 00:14:54,955
This is not quite what I want to
be doing. It was too focused on

243
00:14:54,955 --> 00:14:58,711
one specific thing So I called
Brian up and asked him if the

244
00:14:58,711 --> 00:15:02,405
offer was still open. He said
yeah, and now we do remote as

245
00:15:02,405 --> 00:15:06,162
well And so I ended up working
for Open Robotics Remote from

246
00:15:06,162 --> 00:15:09,856
Japan. Yeah, and then just
before the acquisition happened,

247
00:15:09,856 --> 00:15:13,612
I was called up and said, hey,
this is happening But we want

248
00:15:13,612 --> 00:15:17,306
you to go to the foundation and
help build it up to be much

249
00:15:17,306 --> 00:15:21,000
stronger for the community It's
a winding path really Yeah,

250
00:15:21,000 --> 00:15:24,373
Audrow Nash: Yeah. And we'll be
talking about the acquisition

251
00:15:24,373 --> 00:15:27,912
lots in the future. So that's a
teaser for what's coming. but so

252
00:15:27,912 --> 00:15:31,452
Tully, tell me about your path.
Cause you've been involved right

253
00:15:31,452 --> 00:15:33,000
from the beginning, I think.

254
00:15:33,000 --> 00:15:37,133
Tully Foote: Yeah, I, was
involved with ROS before ROS had

255
00:15:37,133 --> 00:15:41,550
a name or the current code base.
We had early prototypes early

256
00:15:41,550 --> 00:15:45,541
on. Indeed, we started off in
subversion and part of our

257
00:15:45,541 --> 00:15:49,816
messaging early on is that like
we were going to be a proper

258
00:15:49,816 --> 00:15:54,162
open source project and do it on
SourceForge, if you remember

259
00:15:54,162 --> 00:15:58,580
that from way back in the day.
We were there for quite a while

260
00:15:58,580 --> 00:16:02,000
at Willow Garage doing all our
early prototypes.

261
00:16:02,000 --> 00:16:12,510
Geoff Biggs: if you look on
SourceForge for I think it's a

262
00:16:12,510 --> 00:16:23,201
personal robot something or
other, you'll probably find the

263
00:16:23,201 --> 00:16:29,000
old project still sitting there.

264
00:16:29,000 --> 00:16:33,412
Tully Foote: and once we started
getting hey, we think this is

265
00:16:33,412 --> 00:16:37,540
the right one, I was asked to
Build the first test library

266
00:16:37,540 --> 00:16:41,739
where we would be able to,
validate that ROS seemed to work

267
00:16:41,739 --> 00:16:46,294
as a structure and that somebody
could build a library on top of

268
00:16:46,294 --> 00:16:50,564
it and from my history in the
car space and keeping track of

269
00:16:50,564 --> 00:16:54,407
localization and everything,
transforms are always the

270
00:16:54,407 --> 00:16:58,819
biggest challenge. And so what I
did was I wrote the transform

271
00:16:58,819 --> 00:17:03,232
library as a proof of concept
that we can have a library. That

272
00:17:03,232 --> 00:17:07,786
could let be leveraged the power
of ROS in a distributed system.

273
00:17:07,786 --> 00:17:08,000
And

274
00:17:08,000 --> 00:17:11,000
Audrow Nash: So you wrote the
original TF library?

275
00:17:11,000 --> 00:17:12,000
Tully Foote: Yeah.

276
00:17:12,000 --> 00:17:13,000
Audrow Nash: It's awesome. feel

277
00:17:13,000 --> 00:17:15,000
Tully Foote: so from there I,
sorry, go

278
00:17:15,000 --> 00:17:18,818
Audrow Nash: I've been involved
in Open Robotics and now

279
00:17:18,818 --> 00:17:23,045
Intrinsic for a few years, but,
and obviously like I've worked

280
00:17:23,045 --> 00:17:27,068
very closely with Tully, and
Geoff, we bump into each other

281
00:17:27,068 --> 00:17:31,227
pretty often too, but it's so
cool to hear your, both of your

282
00:17:31,227 --> 00:17:35,523
paths. cause I didn't know a lot
of this about you Tully. sorry

283
00:17:35,523 --> 00:17:39,000
to interrupt it, but keep going.
It's just so cool.

284
00:17:39,000 --> 00:17:43,245
Tully Foote: Yeah. so I started
off there writing the libraries,

285
00:17:43,245 --> 00:17:47,158
continued working, did more,
started, got into more release

286
00:17:47,158 --> 00:17:51,005
management, did some management,
wore lots of hats at Open

287
00:17:51,005 --> 00:17:52,000
Robotics. Yeah.

288
00:17:52,000 --> 00:17:56,566
Audrow Nash: one thing that'd be
an interesting dimension is like

289
00:17:56,566 --> 00:18:00,570
the hat you were wearing and
roughly the number of people

290
00:18:00,570 --> 00:18:04,364
that were involved, at Open
Robotics, like So when you

291
00:18:04,364 --> 00:18:08,579
started getting into releases,
was it 10 people, 15, 20, who

292
00:18:08,579 --> 00:18:09,000
knows?

293
00:18:09,000 --> 00:18:13,766
Tully Foote: Oh, like the
release management started off

294
00:18:13,766 --> 00:18:19,043
back at, Willow Garage. I was
the ROS release manager. I think

295
00:18:19,043 --> 00:18:21,000
my first one was Groovy

296
00:18:21,000 --> 00:18:22,000
Audrow Nash: what'd you say,

297
00:18:22,000 --> 00:18:23,000
Tully Foote: we

298
00:18:22,000 --> 00:18:26,366
Geoff Biggs: Willow Garage had
about 60 full time engineers as

299
00:18:26,366 --> 00:18:27,000
I recall.

300
00:18:27,000 --> 00:18:28,000
Tully Foote: yeah,

301
00:18:28,000 --> 00:18:29,000
Audrow Nash: Did they all?

302
00:18:29,000 --> 00:18:30,000
Geoff Biggs: quite a big team.

303
00:18:30,000 --> 00:18:32,360
Tully Foote: about 60 full time
engineers and about 40 peak

304
00:18:32,360 --> 00:18:34,800
contractors and interns in the
building at any given time. So

305
00:18:34,800 --> 00:18:36,000
it was about a hundred people.

306
00:18:36,000 --> 00:18:37,000
Audrow Nash: fun.

307
00:18:37,000 --> 00:18:41,000
Geoff Biggs: Plus the chef.
Don't forget the chef.

308
00:18:41,000 --> 00:18:42,000
Audrow Nash: yeah, I

309
00:18:42,000 --> 00:18:45,000
Geoff Biggs: a full time chef on
staff at Willow Garage.

310
00:18:45,000 --> 00:18:48,000
Audrow Nash: oh, a chef. Oh man.
I wish

311
00:18:48,000 --> 00:18:50,000
Geoff Biggs: food. It was really
good food.

312
00:18:50,000 --> 00:18:53,958
Audrow Nash: I missed this. I, I
was a little too late in

313
00:18:53,958 --> 00:18:58,125
everything and I missed the
great Willow Garage Days. so now

314
00:18:58,125 --> 00:19:02,639
I get to hear about them through
these interviews nonstop. but it

315
00:19:02,639 --> 00:19:07,014
sounds amazing. And there was a
chef. That's great. okay. Sixty

316
00:19:07,014 --> 00:19:10,972
people at Willow Garage,
roughly, plus a bunch of interns

317
00:19:10,972 --> 00:19:15,139
and things like this. And then
how many people came to, Open

318
00:19:15,139 --> 00:19:19,236
Robotics with Brian, when this,
so Tully, I assume you were

319
00:19:19,236 --> 00:19:23,681
there. Was it just a handful or
was it a large part of the group

320
00:19:23,681 --> 00:19:25,000
or how would, how'd

321
00:19:25,000 --> 00:19:28,628
Tully Foote: it started
moderately small. So I think

322
00:19:28,628 --> 00:19:32,814
just for clarification though,
the ROS team at Willow Garage

323
00:19:32,814 --> 00:19:37,279
specifically was only about six
or seven people. They're a large

324
00:19:37,279 --> 00:19:41,465
fraction of Willow Garage was
actually really focused on the

325
00:19:41,465 --> 00:19:45,581
research side and building out a
lot more of the high level

326
00:19:45,581 --> 00:19:49,488
tools, high level research, PR2
specific things. but the

327
00:19:49,488 --> 00:19:53,465
foundation got started with.
It's first contract was with

328
00:19:53,465 --> 00:19:57,512
DARPA to support the DARPA
Robotics Challenge and the team

329
00:19:57,512 --> 00:19:58,000
started

330
00:19:58,000 --> 00:20:00,000
Audrow Nash: vehicle one.
Correct.

331
00:20:00,000 --> 00:20:02,000
Tully Foote: this is the
humanoid walking one.

332
00:20:02,000 --> 00:20:03,000
Audrow Nash: Oh, the humanoid
walking one. Okay.

333
00:20:03,000 --> 00:20:07,911
Geoff Biggs: Yeah, this is after
Fukushima, and so DARPA had a

334
00:20:07,911 --> 00:20:11,000
big thing about humanizing for
disaster

335
00:20:11,000 --> 00:20:15,099
Tully Foote: that was the main
project that kicked off the

336
00:20:15,099 --> 00:20:19,269
foundation and it actually
started as, just Gazebo. The ROS

337
00:20:19,269 --> 00:20:23,367
team was originally, stayed at
Willow Garage. a little bit

338
00:20:23,367 --> 00:20:26,830
later, the foundation got
funding and three of us

339
00:20:26,830 --> 00:20:31,000
transitioned over. myself, Dirk
Thomas and William Woodall.

340
00:20:31,000 --> 00:20:35,225
Audrow Nash: Okay, so cool to
hear about this, because yeah,

341
00:20:35,225 --> 00:20:39,662
Dirk and William, so cool to see
that they've been contributing

342
00:20:39,662 --> 00:20:41,000
like the whole time

343
00:20:41,000 --> 00:20:42,000
Tully Foote: Yep.

344
00:20:42,000 --> 00:20:45,544
Audrow Nash: Okay, so you got
some initial funding, a few

345
00:20:45,544 --> 00:20:49,337
people came over, you, Dirk and
William, and I suppose it was

346
00:20:49,337 --> 00:20:53,254
Brian already, so it was, like
roughly you four, maybe a couple

347
00:20:53,254 --> 00:20:54,000
other folks?

348
00:20:54,000 --> 00:20:59,021
Tully Foote: I think at that
time the gazebo team was about

349
00:20:59,021 --> 00:21:03,872
six or seven already. they
started building up to support

350
00:21:03,872 --> 00:21:08,809
the robotics challenge. so as we
came in as like a smaller

351
00:21:08,809 --> 00:21:14,000
portion of a slightly bigger
organization, but total, like 10

352
00:21:14,000 --> 00:21:19,299
Audrow Nash: You were helping
actively with the development of

353
00:21:19,299 --> 00:21:24,342
ROS, and then you wrote the TF
library, so proving that you

354
00:21:24,342 --> 00:21:29,385
could build libraries on top of
it, and then, how do we, so

355
00:21:29,385 --> 00:21:33,915
then, how many years went by?
How many years was Open

356
00:21:33,915 --> 00:21:38,786
Robotics, or I guess, how many
years was, did it have the

357
00:21:38,786 --> 00:21:43,487
commercial component of it?
Cause clearly OSRF is still

358
00:21:43,487 --> 00:21:44,000
going.

359
00:21:44,000 --> 00:21:48,177
Tully Foote: Yeah. the
foundation got started in 2012.

360
00:21:48,177 --> 00:21:52,972
It's still going. I can't say
off the top of my head when OSRC

361
00:21:52,972 --> 00:21:57,304
got incorporated and we
restructured, I want to say that

362
00:21:57,304 --> 00:21:58,000
was only,

363
00:21:58,000 --> 00:22:03,083
Geoff Biggs: 2013, I think. I
remember Brian telling me about

364
00:22:03,083 --> 00:22:08,000
it at ICRA in Anchorage, which
I'm pretty sure it was 2013.

365
00:22:08,000 --> 00:22:09,000
Audrow Nash: Cool.

366
00:22:09,000 --> 00:22:12,636
Geoff Biggs: Yeah. It was a
while. it was around for quite a

367
00:22:12,636 --> 00:22:13,000
while.

368
00:22:13,000 --> 00:22:18,730
Audrow Nash: 10 years or so
before the acquisition. Okay.

369
00:22:18,730 --> 00:22:24,862
And then Tully, fast forwarding
through the evolution of Open

370
00:22:24,862 --> 00:22:31,095
Robotics. what were some of the
larger milestones between 2013

371
00:22:31,095 --> 00:22:32,000
and 2023?

372
00:22:32,000 --> 00:22:36,762
Tully Foote: Especially for ROS
was the release of ROS 2, Willow

373
00:22:36,762 --> 00:22:41,301
Garage and that legacy was very
focused on ROS 1. And when we

374
00:22:41,301 --> 00:22:45,691
started at Open Robotics, one of
the things that was really

375
00:22:45,691 --> 00:22:50,007
foundational was, we'd ran a
workshop in early 2013, which

376
00:22:50,007 --> 00:22:54,546
was called ROS for Products. And
we talked to a lot of people

377
00:22:54,546 --> 00:22:59,085
about how can we take ROS from
the research field which we've

378
00:22:59,085 --> 00:23:03,922
been very focused on in ROS 1 to
commercial space. And there were

379
00:23:03,922 --> 00:23:07,568
a lot of things that we
identified, such as cross

380
00:23:07,568 --> 00:23:12,033
platform, embedded targets,
installation targets, modularity

381
00:23:12,033 --> 00:23:16,348
and, reliability in terms of
determinism for a startup and

382
00:23:16,348 --> 00:23:21,036
things like that. And these were
the things that we talked to a

383
00:23:21,036 --> 00:23:25,724
bunch of product managers. They,
we had people from NASA there.

384
00:23:25,724 --> 00:23:30,114
We had people from autonomous
driving, lots of people using

385
00:23:30,114 --> 00:23:34,505
ROS and wanted to use ROS in
different spaces. but a lot of

386
00:23:34,505 --> 00:23:38,746
times our collaborators at Bosch
prototyped an autonomous

387
00:23:38,746 --> 00:23:43,434
lawnmower with ROS at the Palo
Alto research facility, and they

388
00:23:43,434 --> 00:23:47,824
got it green lit for being
turned into a product, which you

389
00:23:47,824 --> 00:23:52,437
can now buy. but the first thing
that the product team did was

390
00:23:52,437 --> 00:23:56,902
throw out all of the research
code written in ROS 1, because

391
00:23:56,902 --> 00:24:01,739
they weren't able to trust it to
work. The research teams trusted

392
00:24:01,739 --> 00:24:06,203
it, but the products teams did
not. And we wanted to build a

393
00:24:06,203 --> 00:24:10,370
new version of ROS, which
product teams could trust. And

394
00:24:10,370 --> 00:24:14,760
that was the impetus for ROS 2.
And so at Open Robotics, we

395
00:24:14,760 --> 00:24:19,225
basically took that from the
ground and built it all the way

396
00:24:19,225 --> 00:24:23,615
up to ROS 2, which was released.
Several years ago now, and

397
00:24:23,615 --> 00:24:27,931
that's now what we consider
mainline, and we're being used

398
00:24:27,931 --> 00:24:32,544
in products all over the world.
The, biggest deployment is all

399
00:24:32,544 --> 00:24:35,000
Roombas newer than i7, I
believe.

400
00:24:35,000 --> 00:24:37,000
Audrow Nash: Hell yeah. I have
one in my

401
00:24:37,000 --> 00:24:39,000
Tully Foote: millions of robots.

402
00:24:39,000 --> 00:24:43,295
Audrow Nash: It's so cool. It's
so interesting. So we had ROS 1.

403
00:24:43,295 --> 00:24:47,456
ROS 1 came out of Player, which
came out of, WillowGarage. Was

404
00:24:47,456 --> 00:24:49,000
Player at WillowGarage?

405
00:24:49,000 --> 00:24:52,676
Geoff Biggs: There's no direct
connection between Player and

406
00:24:52,676 --> 00:24:56,291
ROS 1. ROS 1 grew out of a
project called Switchyard, which

407
00:24:56,291 --> 00:24:59,845
was done at Stanford. It's a way
to try and improve on the

408
00:24:59,845 --> 00:25:03,643
concepts of Player as part of
its goals. but, there's a lot of

409
00:25:03,643 --> 00:25:07,381
intellectual legacy because a
lot of the people who worked on

410
00:25:07,381 --> 00:25:11,241
player also worked on ROS, and,
Brian was the leader of both of

411
00:25:11,241 --> 00:25:15,039
them and so on, It's all mixed
up, but there's no direct code.

412
00:25:15,039 --> 00:25:17,000
Legacy that I'm aware of anyway.

413
00:25:17,000 --> 00:25:21,493
Audrow Nash: Makes sense. okay,
but we had ROS 1 and ROS 1 was

414
00:25:21,493 --> 00:25:25,333
heavily used in the research
world, but then now Open

415
00:25:25,333 --> 00:25:29,536
Robotics gets up and running, it
gets funded, it's getting

416
00:25:29,536 --> 00:25:33,884
contract work, and then ROS 2
comes along as a way of making

417
00:25:33,884 --> 00:25:38,594
it so that we can bridge the gap
from research to production. And

418
00:25:38,594 --> 00:25:42,000
that was a big step for the
robotics community.

419
00:25:42,000 --> 00:25:46,261
Audrow Nash: I think. this
probably starts going towards

420
00:25:46,261 --> 00:25:50,750
the robot revolution that you
were mentioning, Geoff. Would

421
00:25:50,750 --> 00:25:55,315
you tell me a bit about it and
how it's, I dunno why this is

422
00:25:55,315 --> 00:25:56,000
exciting,

423
00:25:56,000 --> 00:25:59,584
Geoff Biggs: Back when I did my
PhD and then when I started as a

424
00:25:59,584 --> 00:26:02,889
researcher, service robots,
robots outside of factories and

425
00:26:02,889 --> 00:26:06,249
outside of cages was still very
much like a dream. It's like

426
00:26:06,249 --> 00:26:09,554
this is, maybe we can do it in
really simple cases like the

427
00:26:09,554 --> 00:26:12,802
Roomba, but anything more
complex than that, we don't have

428
00:26:12,802 --> 00:26:16,275
the computing power, we don't
have the software, it's too hard

429
00:26:16,275 --> 00:26:19,579
to develop. and ROS really
solved the too hard to developer

430
00:26:19,579 --> 00:26:22,827
thing, pretty much. And ROS
offered three important things

431
00:26:22,827 --> 00:26:26,244
to people building robots who
wanted to build complex service

432
00:26:26,244 --> 00:26:29,380
robots. It offered, first of
all, it offered really good

433
00:26:29,380 --> 00:26:32,909
developer tools. Something that
all the actors before it didn't

434
00:26:32,909 --> 00:26:36,045
do, including Player. They
didn't offer developer tools.

435
00:26:36,045 --> 00:26:39,294
They didn't offer things like
RViz or ROS bag. These tools

436
00:26:39,294 --> 00:26:42,654
that just make your life easier.
You can understand what the

437
00:26:42,654 --> 00:26:45,902
robot's seeing, for example. It
also offered functionality

438
00:26:45,902 --> 00:26:49,039
libraries, the TF library. Very
important one there. The

439
00:26:49,039 --> 00:26:52,679
navigation stack for ROSFarm was
very helpful for prototyping and

440
00:26:52,679 --> 00:26:56,040
getting things off the ground
and so on. And the third thing

441
00:26:56,040 --> 00:26:59,400
was it made it a lot easier for
people to exchange software.

442
00:26:59,400 --> 00:27:03,041
even in Player it was hard to do
that because everything you used

443
00:27:03,041 --> 00:27:06,569
with Player had to be built into
Player. so it was very hard to

444
00:27:06,569 --> 00:27:09,818
add new stuff to Player, which
was annoying. But with ROS,

445
00:27:09,818 --> 00:27:13,402
everyone could distribute a new
package. And it just worked with

446
00:27:13,402 --> 00:27:16,818
all the other stuff because all
the interfaces were the same.

447
00:27:16,818 --> 00:27:20,123
And because of those three
things, it suddenly became a lot

448
00:27:20,123 --> 00:27:23,651
faster for people to prototype
and then iterate and improve the

449
00:27:23,651 --> 00:27:27,180
robots they were building to the
point where you could actually

450
00:27:27,180 --> 00:27:30,540
have useful services. And so
companies like Fetch, they took

451
00:27:30,540 --> 00:27:33,901
this and they built on it and
actually built usable products

452
00:27:33,901 --> 00:27:37,485
and did the product engineering
to make sure that underneath was

453
00:27:37,485 --> 00:27:41,126
ROS doing all the hard stuff and
they had a nice interface on top

454
00:27:41,126 --> 00:27:44,598
for the end user to use. And,
suddenly everyone's doing robots

455
00:27:44,598 --> 00:27:48,070
everywhere. And this is just in
the past 10 years, it's really

456
00:27:48,070 --> 00:27:51,599
exploded the number of robotic
startups. And now we see them on

457
00:27:51,599 --> 00:27:54,287
farms, we see them in
warehouses, we see them in

458
00:27:54,287 --> 00:27:57,592
Antarctica, they're everywhere,
right? So it's really quite

459
00:27:57,592 --> 00:28:00,000
amazing. It's just like
Exponential growth.

460
00:28:00,000 --> 00:28:04,476
Audrow Nash: For sure. I thought
one fact that Melanie brought up

461
00:28:04,476 --> 00:28:08,332
in our interview, a few back
that Amazon has 750, 000 of

462
00:28:08,332 --> 00:28:12,257
their Kiva robots, the ones that
bring the shelves to the

463
00:28:12,257 --> 00:28:16,251
pickers, to the people who pick
from them, that's bonkers.

464
00:28:16,251 --> 00:28:20,521
That's so many robots. And then
the number of Roombas that are

465
00:28:20,521 --> 00:28:23,000
out there, like it's, very
exciting.

466
00:28:23,000 --> 00:28:26,813
Geoff Biggs: Yeah. And you think
that 20 years ago, the first

467
00:28:26,813 --> 00:28:30,375
Roomba had only just been
released and Kiva didn't exist.

468
00:28:30,375 --> 00:28:34,188
And it's in two decades, we've
just skyrocketed the number of

469
00:28:34,188 --> 00:28:38,188
robots being used in the world.
It's incredible. And, and things

470
00:28:38,188 --> 00:28:41,250
like Dusty, with their
construction robots, these

471
00:28:41,250 --> 00:28:44,813
things just weren't possible
even 10 years ago. And then,

472
00:28:44,813 --> 00:28:48,375
yeah, ROS just changed the
equation because it just came,

473
00:28:48,375 --> 00:28:52,000
it became economically possible
to develop a complex robot

474
00:28:52,000 --> 00:28:55,000
application. And a lot of that
is thanks to ROS.

475
00:28:55,000 --> 00:28:58,824
Audrow Nash: I think so too.
Hell yeah. Okay, so we've talked

476
00:28:58,824 --> 00:29:02,146
a lot about ROS, but Open
Robotics was maintaining or

477
00:29:02,146 --> 00:29:05,970
stewarding other products too.
Tully, would you tell me about

478
00:29:05,970 --> 00:29:09,857
all of the main projects at Open
Robotics? Or actually, if you

479
00:29:09,857 --> 00:29:13,555
guys want to alternate, that'd
be cool too. other than ROS,

480
00:29:13,555 --> 00:29:16,000
what else did we have at Open
Robotics?

481
00:29:16,000 --> 00:29:19,698
Tully Foote: Sure. So the other
one we've talked about is

482
00:29:19,698 --> 00:29:23,915
Gazebo. That's the, that was the
one that I mentioned. It was the

483
00:29:23,915 --> 00:29:27,808
first project that got kicked
off within Open Robotics. This

484
00:29:27,808 --> 00:29:31,376
also has legacy back to the
player era. Brian and other

485
00:29:31,376 --> 00:29:35,528
collaborators developed Player.
And then they realized that they

486
00:29:35,528 --> 00:29:39,161
really wanted to be able to test
this. And so some other

487
00:29:39,161 --> 00:29:43,054
colleagues developed Stage,
which is a 2D simulator. For the

488
00:29:43,054 --> 00:29:45,000
test, the robot agents around.

489
00:29:45,000 --> 00:29:48,688
Geoff Biggs: It was actually
round the other way. Stage was

490
00:29:48,688 --> 00:29:52,563
first. Richard Vaughan, who was
a lab mate of Brian Gerkey, he

491
00:29:52,563 --> 00:29:56,125
developed a simulator called
Arena. And then they decided

492
00:29:56,125 --> 00:29:59,688
they wanted to run their
software on the simulator and on

493
00:29:59,688 --> 00:30:03,563
the robot, which was the first
impetus for Player. And because

494
00:30:03,563 --> 00:30:07,125
they called it Player from the
Shakespeare quote, all the

495
00:30:07,125 --> 00:30:11,000
world's a stage and all people
are only players or whatever it

496
00:30:11,000 --> 00:30:14,000
is, they called, renamed the
simulator to Arena.

497
00:30:14,000 --> 00:30:15,000
Audrow Nash: I had no idea!

498
00:30:15,000 --> 00:30:19,074
Geoff Biggs: A couple of years
later, Nate Koenig came along

499
00:30:19,074 --> 00:30:23,284
and he said, we could do a 3D
one now and because, what's a 3D

500
00:30:23,284 --> 00:30:26,000
stage? It's a gazebo. There's
your name.

501
00:30:26,000 --> 00:30:27,000
Audrow Nash: Wonderful.

502
00:30:27,000 --> 00:30:30,000
Geoff Biggs: it all goes right
back to the beginning. Yeah,

503
00:30:30,000 --> 00:30:32,000
Audrow Nash: Oh, yeah. Okay, so
we have Gazebo

504
00:30:32,000 --> 00:30:34,363
Geoff Biggs: is actually still
around as well you can still use

505
00:30:34,363 --> 00:30:35,000
it today with ROS

506
00:30:35,000 --> 00:30:36,000
Audrow Nash: Wow.

507
00:30:36,000 --> 00:30:37,000
Geoff Biggs: very

508
00:30:37,000 --> 00:30:41,375
Tully Foote: Going back, Gazebo
was this 3D simulator that had

509
00:30:41,375 --> 00:30:45,820
this older, longer legacy, and
we picked it up at Willow Garage

510
00:30:45,820 --> 00:30:49,701
and basically brought it in and
made it the first class

511
00:30:49,701 --> 00:30:53,653
simulator, for the PR2 usage,
and to be the tool that we

512
00:30:53,653 --> 00:30:57,957
wanted the ROS community to be
able to use and leverage. it's

513
00:30:57,957 --> 00:31:01,697
not the only simulator
available, but this is the one

514
00:31:01,697 --> 00:31:05,648
that we've put the most effort
in and have the strongest

515
00:31:05,648 --> 00:31:10,094
integrations with. And so with
that, Gazebo works independently

516
00:31:10,094 --> 00:31:13,622
of ROS but also is tightly
integrated through well

517
00:31:13,622 --> 00:31:17,856
developed plugins. And we've
pushed it, it obviously had its

518
00:31:17,856 --> 00:31:22,089
beginnings back in the Player
Stage era, where it was little

519
00:31:22,089 --> 00:31:26,182
ground rope differential drive
robots running around, some

520
00:31:26,182 --> 00:31:30,486
maybe holonomic ones. We pushed
it up, it did some, it's done

521
00:31:30,486 --> 00:31:34,861
autonomous car simulation for
the DARPA Robotics Challenge. We

522
00:31:34,861 --> 00:31:39,307
built out its capabilities about
doing simulated humanoids. And

523
00:31:39,307 --> 00:31:43,540
walking to the level that we can
run the same motion control

524
00:31:43,540 --> 00:31:47,703
primitives on the simulated
humanoid robot as the live one,

525
00:31:47,703 --> 00:31:51,867
and it continues to walk and
function. With other projects,

526
00:31:51,867 --> 00:31:56,312
we've pushed Gazebo to support
flying vehicles, both fixed wing

527
00:31:56,312 --> 00:31:59,911
and multi propeller ones, using
first level physics

528
00:31:59,911 --> 00:32:04,427
approximations of lift and drag.
we've then taken that below the

529
00:32:04,427 --> 00:32:08,237
ocean and done underwater
vehicles to support offshore

530
00:32:08,237 --> 00:32:12,471
research and testing because it
turns out underwater is very

531
00:32:12,471 --> 00:32:16,916
hard to test. And so simulation
is very important because it's,

532
00:32:16,916 --> 00:32:21,362
you can't communicate with your
underwater, survey device until

533
00:32:21,362 --> 00:32:25,736
it resurfaces. And you really
want to make sure that you don't

534
00:32:25,736 --> 00:32:30,182
have a bug in your code that up
is down and down is up. And you

535
00:32:30,182 --> 00:32:34,768
send your robot to the bottom of
the ocean. All of them have very

536
00:32:34,768 --> 00:32:38,791
fancy fail safe mechanisms to
avoid that, but if you can,

537
00:32:38,791 --> 00:32:43,095
every, any single bug like that
would cause a mission to fail

538
00:32:43,095 --> 00:32:45,000
can save you weeks of work.

539
00:32:45,000 --> 00:32:48,675
Audrow Nash: And like hundreds
of thousands of dollars too, I

540
00:32:48,675 --> 00:32:50,000
imagine. I interviewed

541
00:32:50,000 --> 00:32:52,000
Tully Foote: Ship time is very
expensive.

542
00:32:52,000 --> 00:32:55,962
Audrow Nash: For sure. You're
talking about MBARI? For those

543
00:32:55,962 --> 00:33:00,057
interested, I interviewed them
on the Sense Think Act Podcast,

544
00:33:00,057 --> 00:33:04,217
Ben from MBARI, and that is cool
because their robots are going

545
00:33:04,217 --> 00:33:06,000
to sea for weeks at a time.

546
00:33:06,000 --> 00:33:23,000
Geoff Biggs: of them go for
months. Amazing.

547
00:33:23,000 --> 00:33:24,000
Tully Foote: ROS.

548
00:33:24,000 --> 00:33:27,824
Audrow Nash: hell yeah. Let's
see. And then from talking to

549
00:33:27,824 --> 00:33:31,583
another interview with Louise,
who was leading Gazebo, the

550
00:33:31,583 --> 00:33:35,667
Gazebo team at Open Robotics for
a while, she was saying one of

551
00:33:35,667 --> 00:33:39,815
the big advantages of Gazebo is
that it's really modular. And so

552
00:33:39,815 --> 00:33:43,963
if you are simulating something
simple, but wanted to run really

553
00:33:43,963 --> 00:33:47,722
fast, you can disable or just
not include all of the parts

554
00:33:47,722 --> 00:33:51,611
that you're not using in your
simulation and then because of

555
00:33:51,611 --> 00:33:55,694
that you can run at super fast
speeds just because you're doing

556
00:33:55,694 --> 00:33:59,907
way less work, whereas it's very
hard with completely opaque, not

557
00:33:59,907 --> 00:34:04,056
open source simulators and this
kind of thing. So I thought that

558
00:34:04,056 --> 00:34:06,000
was a big advantage of Gazebo.

559
00:34:06,000 --> 00:34:09,526
Tully Foote: Yeah, the physics
engine is swappable, the

560
00:34:09,526 --> 00:34:13,115
rendering engine is swappable,
so people have plugged in

561
00:34:13,115 --> 00:34:16,449
various, hardware accelerated
renderers and hardware

562
00:34:16,449 --> 00:34:20,551
accelerated physics engines for
different research projects, and

563
00:34:20,551 --> 00:34:24,269
we can, let that happen. And
the, super fast one is called

564
00:34:24,269 --> 00:34:26,000
the Trivial Physics Engine.

565
00:34:26,000 --> 00:34:28,357
Geoff Biggs:
SimplePhysicsEngine. Yeah. it's

566
00:34:28,357 --> 00:34:29,000
very simple.

567
00:34:29,000 --> 00:34:30,000
Audrow Nash: similar. Yeah.

568
00:34:30,000 --> 00:34:31,000
Geoff Biggs: Yeah.

569
00:34:31,000 --> 00:34:32,000
Audrow Nash: That's funny

570
00:34:32,000 --> 00:34:36,000
Geoff Biggs: It is trivial. It's
just very basic physics.

571
00:34:36,000 --> 00:34:38,769
Audrow Nash: Hell yeah. That's
great. Sometimes it's all you

572
00:34:38,769 --> 00:34:39,000
need.

573
00:34:39,000 --> 00:34:40,000
Geoff Biggs: It is.

574
00:34:39,000 --> 00:34:42,544
Geoff Biggs: We actually, Open
Robotics used it heavily for the

575
00:34:42,544 --> 00:34:45,975
OpenRMF project, because that
was fleet management. we didn't

576
00:34:45,975 --> 00:34:49,350
really care if the robots were
being accurately simulated in

577
00:34:49,350 --> 00:34:52,838
terms of avoiding obstacles and
real grip and all that sort of

578
00:34:52,838 --> 00:34:56,156
stuff. Just wanted to know they
were moving around on their

579
00:34:56,156 --> 00:34:57,000
paths properly.

580
00:34:57,000 --> 00:35:00,539
Audrow Nash: So actually that's
a good segue, to Geoff, tell me

581
00:35:00,539 --> 00:35:02,000
about the OpenRMF project.

582
00:35:02,000 --> 00:35:04,980
Geoff Biggs: Yeah, so the
OpenRMF project started back in

583
00:35:04,980 --> 00:35:08,275
2018, I think it was. And this
was actually in, A proposal from

584
00:35:08,275 --> 00:35:11,098
the government of Singapore,
their research arm, and a

585
00:35:11,098 --> 00:35:14,497
hospital in Singapore. They came
to Open Robotics and said, we've

586
00:35:14,497 --> 00:35:17,529
got this problem. Our hospital
has got all these robots in

587
00:35:17,529 --> 00:35:20,510
them, but they come from
different manufacturers, and all

588
00:35:20,510 --> 00:35:23,752
the fleets don't talk to each
other. And you get two different

589
00:35:23,752 --> 00:35:26,523
robots from different
manufacturers trying to use the

590
00:35:26,523 --> 00:35:29,765
elevator at the same time, and
they get stuck, and someone has

591
00:35:29,765 --> 00:35:32,850
to go and sort it out. And so
this was like a problem. They

592
00:35:32,850 --> 00:35:34,000
went, yeah, basically.

593
00:35:34,000 --> 00:35:35,000
Audrow Nash: Yeah.

594
00:35:35,000 --> 00:35:38,700
Geoff Biggs: The cause traffic
jams, because they can't talk to

595
00:35:38,700 --> 00:35:42,399
each other at all. And so this
was something that, can you come

596
00:35:42,399 --> 00:35:46,040
and help us figure out how to
solve this? And so Open Robotics

597
00:35:46,040 --> 00:35:49,681
set up the Singapore office for
its initial reason was to work

598
00:35:49,681 --> 00:35:53,322
on that project and that project
turned into It's like a fleet

599
00:35:53,322 --> 00:35:57,022
manager manager basically. The
idea is that it works with fleet

600
00:35:57,022 --> 00:36:00,546
managers to coordinate actions
of robots that can't directly

601
00:36:00,546 --> 00:36:04,187
talk to each other and then that
grew even bigger to the point

602
00:36:04,187 --> 00:36:07,710
where it becomes basically a
sort of universal fleet manager

603
00:36:07,710 --> 00:36:11,293
and the actual fleet managers
for each individual robot don't

604
00:36:11,293 --> 00:36:14,816
really do much thinking for
themselves they just Control the

605
00:36:14,816 --> 00:36:17,987
robots, and OpenRMF is the one
that is responsible for

606
00:36:17,987 --> 00:36:21,511
assigning tasks, figuring out
who, which robot is, and which

607
00:36:21,511 --> 00:36:25,093
fleet is the one, best one to
handle certain tasks and so on.

608
00:36:25,093 --> 00:36:28,029
And so this was built in
Singapore. to work in the

609
00:36:28,029 --> 00:36:31,553
hospital where it is being used
in real life today to manage

610
00:36:31,553 --> 00:36:35,018
several fleets of robots on an
industrial network in a very

611
00:36:35,018 --> 00:36:38,248
critical environment. It's a
hospital, you've got to do

612
00:36:38,248 --> 00:36:41,712
things right, you've got to be
careful. And these robots do

613
00:36:41,712 --> 00:36:44,649
things like delivering
medicines, delivering food,

614
00:36:44,649 --> 00:36:48,055
cleaning the floors, all that
sort of stuff. And they take

615
00:36:48,055 --> 00:36:51,813
elevators, they do all this cool
stuff. And OpenRMF controls all

616
00:36:51,813 --> 00:36:55,219
of that. It even does things
like Saying, okay, this robot

617
00:36:55,219 --> 00:36:59,037
wants to use the elevator, so it
cools the elevator for the robot

618
00:36:59,037 --> 00:37:02,560
and makes sure it gets to the
right floor, or it opens doors

619
00:37:02,560 --> 00:37:06,084
for the robot and so on. It
coordinates all of that for you.

620
00:37:06,084 --> 00:37:09,490
And so this has really become,
it's become the third major

621
00:37:09,490 --> 00:37:12,778
project of, Open Robotics
because it's, as these robots,

622
00:37:12,778 --> 00:37:16,126
out in the world have grown in
number. Something's got to

623
00:37:16,126 --> 00:37:19,532
manage them all to make sure
they coordinate properly. And

624
00:37:19,532 --> 00:37:21,000
that's what OpenRMF does.

625
00:37:21,000 --> 00:37:24,283
Audrow Nash: Yeah, for sure.
Yeah. I imagine anywhere that

626
00:37:24,283 --> 00:37:27,679
we're going to have a lot of
robots, which if we see more of

627
00:37:27,679 --> 00:37:31,075
a service robots explosion, this
kind of thing will be very,

628
00:37:31,075 --> 00:37:33,000
valuable, I imagine. And it's
hard

629
00:37:33,000 --> 00:37:34,000
Geoff Biggs: Yeah,

630
00:37:34,000 --> 00:37:35,814
Audrow Nash: a especially
interfacing with all these

631
00:37:35,814 --> 00:37:37,000
different makes of robots that
had

632
00:37:37,000 --> 00:37:39,000
Geoff Biggs: it's really hard.

633
00:37:39,000 --> 00:37:44,719
Audrow Nash: to each other. And
then I guess, I don't know if

634
00:37:44,719 --> 00:37:48,000
it's worth mentioning, but Tully
is

635
00:37:48,000 --> 00:37:52,000
Geoff Biggs: it's like some of
them are proprietary.

636
00:37:52,000 --> 00:37:54,000
Audrow Nash: Yep.

637
00:37:54,000 --> 00:37:56,465
Geoff Biggs: Some of them are
proprietary, some are open

638
00:37:56,465 --> 00:37:58,798
source, they have different
interfaces to their fleet

639
00:37:58,798 --> 00:38:01,526
managers, the robots themselves
have different interfaces. And

640
00:38:01,526 --> 00:38:04,035
there's so many environments
where this is important, not

641
00:38:04,035 --> 00:38:06,544
just hospitals. Like you can
think of a hotel. They would

642
00:38:06,544 --> 00:38:09,229
have relay robotics come in and
do the delivery, they'll have

643
00:38:09,229 --> 00:38:11,694
iRobot come in and do the
cleaning robots, and all these

644
00:38:11,694 --> 00:38:14,158
things they need to coordinate,
and so on, and all these

645
00:38:14,158 --> 00:38:16,799
different environments, office
buildings, out on the street,

646
00:38:16,799 --> 00:38:19,000
delivery robots. It's really
important everywhere.

647
00:38:19,000 --> 00:38:22,000
Audrow Nash: For sure. Hell
yeah.

648
00:38:22,000 --> 00:38:25,608
Audrow Nash: And then, maybe the
last one to discuss as a

649
00:38:25,608 --> 00:38:29,278
significant project out of Open
Robotics would be like the

650
00:38:29,278 --> 00:38:32,000
infrastructure, the build farm,
things like

651
00:38:32,000 --> 00:38:33,000
Geoff Biggs: Absolutely.

652
00:38:33,000 --> 00:38:35,000
Audrow Nash: Tully, would you
like to speak to this?

653
00:38:35,000 --> 00:38:39,128
Tully Foote: Yeah, so I think
this is one of the things that

654
00:38:39,128 --> 00:38:43,118
is often goes unseen and just,
taken for granted. the Open

655
00:38:43,118 --> 00:38:46,558
Robotics team has done an
amazing job of providing

656
00:38:46,558 --> 00:38:50,479
infrastructure for the entire
worldwide community to take

657
00:38:50,479 --> 00:38:54,745
advantage and make the use of
ROS easy and convenient. you can

658
00:38:54,745 --> 00:38:59,148
go out there and just sudo, add
the ROS sources and sudo apt-get

659
00:38:59,148 --> 00:39:03,482
install the packages on modern,
recent Ubuntu LTSs. And it just

660
00:39:03,482 --> 00:39:07,472
works, and you can be up and
running in minutes. We've got

661
00:39:07,472 --> 00:39:11,738
Docker containers that are pre
built with the latest software.

662
00:39:11,738 --> 00:39:15,866
you can download tarballs and
install them, and you can just

663
00:39:15,866 --> 00:39:20,131
expect them to work. And, we've
got a lot of web services that

664
00:39:20,131 --> 00:39:23,846
are up. The documentation
services, that are there and

665
00:39:23,846 --> 00:39:27,974
available for the community to
take advantage of. And having

666
00:39:27,974 --> 00:39:32,308
all these things there at your
fingertips It's something that a

667
00:39:32,308 --> 00:39:36,711
lot of people are just like, Oh,
it's just there. It just works.

668
00:39:36,711 --> 00:39:40,632
But there's actually a whole
team behind them making sure

669
00:39:40,632 --> 00:39:45,036
that's there and works well and
that has quick response time for

670
00:39:45,036 --> 00:39:49,163
the developer so that you have
the, like, when I make a pull

671
00:39:49,163 --> 00:39:52,672
request against some ROS
package, I'll get a little

672
00:39:52,672 --> 00:39:56,593
response immediately, like
moderately quickly saying Hey,

673
00:39:56,593 --> 00:40:00,859
this passes test, this fails the
test. And then we have deeper

674
00:40:00,859 --> 00:40:05,124
ones that run nightly or that
are manually triggered that will

675
00:40:05,124 --> 00:40:08,771
run cross package, cross
platform tests. So we can be

676
00:40:08,771 --> 00:40:13,242
confident when that gets merged,
it's going to work and you won't

677
00:40:13,242 --> 00:40:16,751
break other people. And as long
as we can keep that

678
00:40:16,751 --> 00:40:20,810
infrastructure and running, it
helps significantly improves

679
00:40:20,810 --> 00:40:24,663
developer speed. Because you
don't have to worry that my

680
00:40:24,663 --> 00:40:28,515
change is actually breaking
things that are in your pull

681
00:40:28,515 --> 00:40:32,781
request later. So that allows us
to keep developer velocity up

682
00:40:32,781 --> 00:40:37,115
both for the core team, but also
for the greater community. The

683
00:40:37,115 --> 00:40:40,899
foundation provides a ton of
resources for building and

684
00:40:40,899 --> 00:40:44,889
releasing packages for the
entire community. If you, as an

685
00:40:44,889 --> 00:40:49,154
individual developer, wanted to
release a ROS package, And you

686
00:40:49,154 --> 00:40:53,351
had to mainly go and build it
and rebuild it every, time your

687
00:40:53,351 --> 00:40:57,204
dependencies were updated. it
would be much, much slower

688
00:40:57,204 --> 00:41:01,263
because you'd have to poke a
whole lot of other people. And

689
00:41:01,263 --> 00:41:04,978
whereas right now the Open
Robotics provides all those

690
00:41:04,978 --> 00:41:09,381
services to the whole community
if they're releasing open source

691
00:41:09,381 --> 00:41:10,000
software.

692
00:41:10,000 --> 00:41:14,443
Audrow Nash: Yeah, for sure. I
think like the build farm is

693
00:41:14,443 --> 00:41:19,338
pretty amazing with all this. So
people will create a ROS package

694
00:41:19,338 --> 00:41:23,932
and then they submit it to us on
the ROS distro website. And,

695
00:41:23,932 --> 00:41:28,375
point to the release repository
and then people can like it

696
00:41:28,375 --> 00:41:32,968
builds it checks it manages all
the dependencies And then you

697
00:41:32,968 --> 00:41:37,713
can as you were saying you can
apt get install it Which is like

698
00:41:37,713 --> 00:41:42,382
other people's packages in the
community and then so this lets

699
00:41:42,382 --> 00:41:46,825
you install big packages like
moveit or nav2 pretty easily,

700
00:41:46,825 --> 00:41:51,117
but even more esoteric ones
still just an apt get install

701
00:41:51,117 --> 00:41:53,000
away, which is quite cool

702
00:41:53,000 --> 00:41:56,681
Tully Foote: Yeah, and I think
that this is something that

703
00:41:56,681 --> 00:42:00,490
modern developers somewhat take
for granted. back in the day

704
00:42:00,490 --> 00:42:04,425
when we had to check everything
out and build from source, and

705
00:42:04,425 --> 00:42:08,234
if there was a bug you had to
stop and SVN check out for the

706
00:42:08,234 --> 00:42:12,169
code and then start the build
and go to lunch and come back to

707
00:42:12,169 --> 00:42:16,295
try it again. I remember when we
first had the binary builds that

708
00:42:16,295 --> 00:42:19,786
are available quick and
moderately quickly, we actually

709
00:42:19,786 --> 00:42:23,531
got to the point that people
were just like, I'm working on

710
00:42:23,531 --> 00:42:26,831
this project. There's a bug
that's going to be fixed

711
00:42:26,831 --> 00:42:30,322
upstream. I'm going to wait the
two days for the, other

712
00:42:30,322 --> 00:42:34,194
maintainer, release, patch, fix,
and build. And then I'll try

713
00:42:34,194 --> 00:42:38,129
again. And I'll go do something
else. and it's a, it's another

714
00:42:38,129 --> 00:42:41,874
whole different way to think
about things when that release

715
00:42:41,874 --> 00:42:45,619
pipeline can be trusted and be
part of your critical supply

716
00:42:45,619 --> 00:42:46,000
chain.

717
00:42:46,000 --> 00:42:49,964
Audrow Nash: That's awesome.
Yeah. And one thing that I've

718
00:42:49,964 --> 00:42:53,996
really liked is there's a site
called index.ros.org. And so

719
00:42:53,996 --> 00:42:58,234
that one, it is an index for all
the packages and it shows you

720
00:42:58,234 --> 00:43:02,335
how they all relate to each
other and which distributions of

721
00:43:02,335 --> 00:43:06,640
ROS they're available for, links
to documentations, link to the

722
00:43:06,640 --> 00:43:10,878
repository. It's a very cool way
to see what all is out there.

723
00:43:10,878 --> 00:43:14,773
And also You can see the
relationships, what packages use

724
00:43:14,773 --> 00:43:18,874
what packages, and just trace
them back really deep. I don't

725
00:43:18,874 --> 00:43:22,701
know. It's a very cool, complex
network that has so many

726
00:43:22,701 --> 00:43:24,000
contributors to it.

727
00:43:24,000 --> 00:43:28,410
Tully Foote: Yes, it's a, that's
a, great website that the

728
00:43:28,410 --> 00:43:33,276
foundation provides and, highly
recommend people. Take advantage

729
00:43:33,276 --> 00:43:37,839
and leverage it. That's one
that, is less, less appreciated.

730
00:43:37,839 --> 00:43:42,173
I think a lot of the content
historically has been on the

731
00:43:42,173 --> 00:43:46,963
wiki, but we are moving to use
index instead for ROS 2, so that

732
00:43:46,963 --> 00:43:51,829
it's a little bit more scalable.
The wiki has reached its limits

733
00:43:51,829 --> 00:43:55,859
and we can't sustain it anymore
due to the underlying

734
00:43:55,859 --> 00:43:57,000
infrastructure.

735
00:43:57,000 --> 00:44:00,000
Audrow Nash: Yeah, that was more
of a ROS 1 thing, and I think

736
00:44:00,000 --> 00:44:04,496
Tully Foote: the MoinMoin wikis
are designed for one to 2000

737
00:44:04,496 --> 00:44:09,217
pages. For the web server and I
believe we're at something like

738
00:44:09,217 --> 00:44:13,339
10 to 20, 000 pages. I have
scripts that go through and

739
00:44:13,339 --> 00:44:17,685
clean up empty pages and spam
pages to clear them out. But

740
00:44:17,685 --> 00:44:22,256
we're still running at something
like 10x the number of pages

741
00:44:22,256 --> 00:44:26,227
that it was designed for and
we've had to throw extra

742
00:44:26,227 --> 00:44:29,000
resources at just to keep it
running.

743
00:44:29,000 --> 00:44:32,261
Geoff Biggs: Yeah, probably the
only reason it still runs is

744
00:44:32,261 --> 00:44:34,000
because CPUs have gotten faster.

745
00:44:34,000 --> 00:44:38,968
Audrow Nash: That's a good
point. Let's see. Okay. So those

746
00:44:38,968 --> 00:44:42,000
are the main areas of Open
Robotics.

747
00:44:42,000 --> 00:44:47,399
Audrow Nash: where were we say
2022 or something like, like

748
00:44:47,399 --> 00:44:52,248
where was Open Robotics? How
were we functioning as a

749
00:44:52,248 --> 00:44:56,000
company? Just where were we?
Maybe Geoff?

750
00:44:56,000 --> 00:44:59,500
Geoff Biggs: I can speak to some
of it. Tully was more involved

751
00:44:59,500 --> 00:45:03,000
in the management at that time.
as a company, Open Robotics was

752
00:45:03,000 --> 00:45:06,611
actually very successful. It was
technically a startup, it's only

753
00:45:06,611 --> 00:45:10,111
10 years old, and it never had
venture funding. It was entirely

754
00:45:10,111 --> 00:45:11,000
bootstrapped. So

755
00:45:11,000 --> 00:45:13,000
Audrow Nash: No debt at all,
right?

756
00:45:13,000 --> 00:45:16,531
Geoff Biggs: Yeah, so for the
company to grow, from just a few

757
00:45:16,531 --> 00:45:20,062
people at the start right up to
about 50 or 60 by the time the

758
00:45:20,062 --> 00:45:23,479
acquisition happened, with no
venture capital is pretty rare

759
00:45:23,479 --> 00:45:27,010
these days, especially in the
technology sector. And it was in

760
00:45:27,010 --> 00:45:30,712
general a pretty healthy company
and I think all three of us here

761
00:45:30,712 --> 00:45:33,958
would absolutely agree it was a
great place to work, very

762
00:45:33,958 --> 00:45:37,603
friendly, very good culture, it
was a wonderful place to work. I

763
00:45:37,603 --> 00:45:40,735
think, I hand over to Tully
though, I think he was more

764
00:45:40,735 --> 00:45:44,095
privy to what was going on at
the management level in those

765
00:45:44,095 --> 00:45:47,000
days. I was trying to be an
engineer at that point.

766
00:45:47,000 --> 00:45:50,863
Tully Foote: Yeah, so I think
that, we'd seen a lot of growth

767
00:45:50,863 --> 00:45:54,980
and we've seen, there's a lot of
progress and things happening in

768
00:45:54,980 --> 00:45:58,589
the space and nearby spaces.
There's a lot of innovations

769
00:45:58,589 --> 00:46:02,453
that are going on that are close
to, in the robotic, space or

770
00:46:02,453 --> 00:46:06,442
close to the robotic space. You
see all the funding for AI, big

771
00:46:06,442 --> 00:46:10,179
companies coming in and doing
big investments. And as Geoff

772
00:46:10,179 --> 00:46:13,979
mentioned, Open Robotics and
OSRC had been bootstrapping all

773
00:46:13,979 --> 00:46:17,588
along. And with our business
model, we couldn't make, big

774
00:46:17,588 --> 00:46:21,388
investments. A lot of what we
did was, find projects that we

775
00:46:21,388 --> 00:46:25,125
would work on that would help
push forward the state of the

776
00:46:25,125 --> 00:46:28,988
art. but we were doing it in a
lot of small steps and finding

777
00:46:28,988 --> 00:46:32,598
ways, to find good customer,
good projects for customers.

778
00:46:32,598 --> 00:46:36,651
That would help us on our vision
for how to improve ROS. We also

779
00:46:36,651 --> 00:46:40,450
wanted to do more strategic
investment and push into regions

780
00:46:40,450 --> 00:46:44,187
that we thought would be, in
bigger directions, take bigger

781
00:46:44,187 --> 00:46:47,607
steps, which is why we looked
for, alternative ways to

782
00:46:47,607 --> 00:46:49,000
structure the project.

783
00:46:49,000 --> 00:46:53,759
Audrow Nash: Hell yeah, so how I
understand it is we had we were

784
00:46:53,759 --> 00:46:57,476
contracting and so from
contracting you get like a

785
00:46:57,476 --> 00:47:01,863
little margin of profit and so
with that profit we had to I

786
00:47:01,863 --> 00:47:06,399
guess, I guess people were paid
through the, I don't know. So

787
00:47:06,399 --> 00:47:10,489
you, get a little profit extra
on top of everything, in

788
00:47:10,489 --> 00:47:14,653
addition to your running
expenses. And with that profit,

789
00:47:14,653 --> 00:47:19,337
you can reinvest it into large
infrastructure changes, big, big

790
00:47:19,337 --> 00:47:23,724
features, larger initiatives,
basically. And so that's very

791
00:47:23,724 --> 00:47:28,185
hard to do on the meager. Not
meager, but like just the thin

792
00:47:28,185 --> 00:47:32,572
layer from contracting, of
profits. And, so because of that

793
00:47:32,572 --> 00:47:37,256
we were very iterative in what
we rolled out. You agree, Tully,

794
00:47:37,256 --> 00:47:38,000
with this?

795
00:47:38,000 --> 00:47:42,239
Tully Foote: I think that's, we
were doing the work and then

796
00:47:42,239 --> 00:47:46,054
reinvesting into the open
source. And we, it's not, we

797
00:47:46,054 --> 00:47:50,364
weren't able to take big steps,
make big moves, invest in big

798
00:47:50,364 --> 00:47:51,000
projects.

799
00:47:51,000 --> 00:47:55,034
Audrow Nash: What, are some big
projects that we'd like to, and

800
00:47:55,034 --> 00:47:59,069
we, what were some ones that we
really wanted to invest in that

801
00:47:59,069 --> 00:48:02,847
would have required some big
steps, versus very incremental

802
00:48:02,847 --> 00:48:04,000
change as we have?

803
00:48:04,000 --> 00:48:08,000
Tully Foote: so the, very
concrete one is the RMW Zenoh.

804
00:48:08,000 --> 00:48:10,000
Audrow Nash: Ah, I'm so excited
about that.

805
00:48:10,000 --> 00:48:14,158
Tully Foote: This is something
that we have, we've seen

806
00:48:14,158 --> 00:48:18,316
potential in Zenoh, but we, and
we'd done some proof of

807
00:48:18,316 --> 00:48:22,625
concepts, and we had some
estimates of what it would take

808
00:48:22,625 --> 00:48:27,464
to do this sort of project. And,
But there was no companies that

809
00:48:27,464 --> 00:48:31,622
were willing to step up and
fund, hey, let's go develop

810
00:48:31,622 --> 00:48:32,000
that.

811
00:48:32,000 --> 00:48:36,138
Audrow Nash: Yep. Just to, for
background with that, we were

812
00:48:36,138 --> 00:48:40,069
using DDS with ROS 2 as the
primary middleware and it has

813
00:48:40,069 --> 00:48:43,931
pretty significant scaling
issues, right? When we have a

814
00:48:43,931 --> 00:48:48,276
lot of nodes, because they have
a lot of inter node traffic, is

815
00:48:48,276 --> 00:48:52,690
it correct? And then Zenoh, it's
another middleware, but it does

816
00:48:52,690 --> 00:48:56,276
things a little bit differently
in a way that scales

817
00:48:56,276 --> 00:49:00,000
significantly better, and it
seems like it's built for

818
00:49:00,000 --> 00:49:04,069
robotics applications or it's
just very well suited for it,

819
00:49:04,069 --> 00:49:08,000
and so thus it will have a lot
fewer middleware problems,

820
00:49:08,000 --> 00:49:11,759
Tully Foote: Yeah, so part of,
what we did with ROS 2 is we

821
00:49:11,759 --> 00:49:15,453
very specifically designed it to
have what we call the ROS

822
00:49:15,453 --> 00:49:19,530
middleware abstraction. And what
this allows us to do is be able

823
00:49:19,530 --> 00:49:22,843
to switch, between our
middleware implementation. we

824
00:49:22,843 --> 00:49:26,984
specifically did this because we
were targeting the DDS standard,

825
00:49:26,984 --> 00:49:30,551
and there were multiple DDS
vendors, and we've, over the

826
00:49:30,551 --> 00:49:34,055
years, we've had different
default vendors. We've had a

827
00:49:34,055 --> 00:49:37,495
whole selection of vendors
available as the middleware

828
00:49:37,495 --> 00:49:41,508
implementation that we used. But
part of what we did was define

829
00:49:41,508 --> 00:49:45,139
that middleware abstraction, and
the nice thing is we can

830
00:49:45,139 --> 00:49:49,025
actually switch to a potentially
non DDS middleware if it can

831
00:49:49,025 --> 00:49:51,000
provide the same functionality.

832
00:49:51,000 --> 00:49:52,000
Audrow Nash: I

833
00:49:52,000 --> 00:49:55,000
Geoff Biggs: Yes again soon.

834
00:49:55,000 --> 00:49:56,000
Tully Foote: there's some

835
00:49:55,000 --> 00:49:58,706
Audrow Nash: middleware. there's
some joke ones that still work

836
00:49:58,706 --> 00:50:00,000
because of abstraction

837
00:50:00,000 --> 00:50:04,103
Tully Foote: we, joked about an
April Fool's one, which would be

838
00:50:04,103 --> 00:50:07,756
a, Airhorn based Morse code
middleware. And we'd have two

839
00:50:07,756 --> 00:50:11,090
robots communicating via
Airhorns across the office.

840
00:50:11,090 --> 00:50:14,487
never got around to that one,
but I do have some like

841
00:50:14,487 --> 00:50:18,654
electrically controlled Airhorns
in the closet if anyone wants to

842
00:50:18,654 --> 00:50:20,000
work with me on that.

843
00:50:20,000 --> 00:50:22,000
Audrow Nash: it.

844
00:50:22,000 --> 00:50:27,000
Geoff Biggs: care of

845
00:50:27,000 --> 00:50:30,706
Tully Foote: I from that one of
the things we identified, we've

846
00:50:30,706 --> 00:50:32,000
done some projects and

847
00:50:32,000 --> 00:50:36,000
Geoff Biggs: that.

848
00:50:36,000 --> 00:50:39,840
Tully Foote: on Zenoh and
there's, there are, definitely

849
00:50:39,840 --> 00:50:44,022
some potential benefits and we
think it's worth investing in.

850
00:50:44,022 --> 00:50:48,273
We've done some projects with
them that have worked out, using

851
00:50:48,273 --> 00:50:52,456
Zenoh that have worked out well
and we want to, we, wanted to

852
00:50:52,456 --> 00:50:56,707
try it out and one of the things
that's like at this scale. We

853
00:50:56,707 --> 00:51:00,821
don't really know how it's going
to work for everybody until

854
00:51:00,821 --> 00:51:05,003
we've actually built it. And so
it's a moderately speculative

855
00:51:05,003 --> 00:51:09,049
thing to go and build this size
of a project and see how it

856
00:51:09,049 --> 00:51:13,231
works. As you mentioned, we've
ran into trouble with a bit of

857
00:51:13,231 --> 00:51:17,414
an impedance mismatch, I think,
between the DDS vision of how

858
00:51:17,414 --> 00:51:21,802
distributed systems work and the
ROS vision. ROS very focused on

859
00:51:21,802 --> 00:51:26,258
nodes and lots of endpoints, and
DDS would really rather do keyed

860
00:51:26,258 --> 00:51:30,715
topics, with filtering on shared
topics. And so we're looking at,

861
00:51:30,715 --> 00:51:34,966
Zenoh as a potential alternative
to try out. And so we're not,

862
00:51:34,966 --> 00:51:39,012
moving away from DDS, but we
want to make the investment to

863
00:51:39,012 --> 00:51:41,000
see how it works, with Zenoh.

864
00:51:41,000 --> 00:51:45,333
Audrow Nash: Hell yeah, and one
thing that's very exciting from

865
00:51:45,333 --> 00:51:49,392
my perspective, for Zenoh is
built in Rust, the programming

866
00:51:49,392 --> 00:51:53,725
language, and Rust, I absolutely
love Rust. I got to use it for

867
00:51:53,725 --> 00:51:57,852
one of my Open Robotics projects
and just completely fell in

868
00:51:57,852 --> 00:52:02,048
love. So all my side projects
are now in Rust. but so because

869
00:52:02,048 --> 00:52:05,899
of that, Rust is going to be
more supported. In the open

870
00:52:05,899 --> 00:52:10,370
robot or that's a big challenge,
I think is to make it so that we

871
00:52:10,370 --> 00:52:14,635
can support Rust and its build
process, which is fundamentally

872
00:52:14,635 --> 00:52:18,556
different to C++ and how it does
versioning and how we do

873
00:52:18,556 --> 00:52:20,000
operating systems and

874
00:52:20,000 --> 00:52:21,000
Geoff Biggs: Yeah,

875
00:52:21,000 --> 00:52:22,000
Audrow Nash: dependencies and

876
00:52:22,000 --> 00:52:25,070
Geoff Biggs: it's been
interesting because, so Rust,

877
00:52:25,070 --> 00:52:28,848
and in particular Cargo and its
way of building things, are very

878
00:52:28,848 --> 00:52:31,917
different from how C++ and
Python and a lot of other

879
00:52:31,917 --> 00:52:35,754
languages we've used up till now
work. And so figuring out how to

880
00:52:35,754 --> 00:52:39,060
integrate that into workspaces
and the Buildfarm, and in

881
00:52:39,060 --> 00:52:42,307
particular, one of the big
problems is that Rust users,

882
00:52:42,307 --> 00:52:45,731
they used to be able to just
rust up the latest version of

883
00:52:45,731 --> 00:52:49,391
Rust, whereas the Buildfarm can
only use what's available from

884
00:52:49,391 --> 00:52:53,169
the system package manager from
the default repositories. And so

885
00:52:53,169 --> 00:52:56,711
that makes it, difficult to say,
what version of Rust do you

886
00:52:56,711 --> 00:53:00,371
support, and so on. But these
are problems that we are working

887
00:53:00,371 --> 00:53:03,972
on solving, and, we do hope to
basically have Rust support in

888
00:53:03,972 --> 00:53:07,809
the Buildfarm. At about the same
time, that we're trying to start

889
00:53:07,809 --> 00:53:11,351
getting Zenoh out there and
then, then things are just gonna

890
00:53:11,351 --> 00:53:12,000
be awesome.

891
00:53:12,000 --> 00:53:15,718
Audrow Nash: Yes, totally. Yeah,
it's funny that there's a

892
00:53:15,718 --> 00:53:19,244
complete mismatch in how they do
dependencies for this.

893
00:53:19,244 --> 00:53:23,218
Completely different paradigms.
And yeah, that's a hard square

894
00:53:23,218 --> 00:53:27,000
to circle, I think. But, okay.
I'm excited about that. But,

895
00:53:27,000 --> 00:53:32,700
Audrow Nash: now we have enter
Intrinsic. I think, maybe Tully,

896
00:53:32,700 --> 00:53:38,038
tell me a bit about Intrinsic
and, the acquisition. I don't

897
00:53:38,038 --> 00:53:43,467
know, just background on all
this and what's occurred. And I

898
00:53:43,467 --> 00:53:46,000
know it was like a year ago,

899
00:53:46,000 --> 00:53:47,000
Tully Foote: Yep. Yeah. It's,

900
00:53:47,000 --> 00:53:48,000
Audrow Nash: Good to figure it
out now.

901
00:53:48,000 --> 00:53:52,292
Tully Foote: bit over a year
ago. we worked out the agreement

902
00:53:52,292 --> 00:53:56,514
to have the open source robotics
corporation, the for profit

903
00:53:56,514 --> 00:54:00,806
subsidiary doing the consulting
work for open robotics. to be

904
00:54:00,806 --> 00:54:05,099
acquired by Intrinsic. this was
a, mutual agreement where we,

905
00:54:05,099 --> 00:54:09,250
believe that it is beneficial
for all the parties involved.

906
00:54:09,250 --> 00:54:13,050
One of the, one of the
interesting things when we were

907
00:54:13,050 --> 00:54:17,483
doing the, had our discussions,
Open Robotics obviously has the

908
00:54:17,483 --> 00:54:21,775
mission to promote open source
robotics, education, outreach,

909
00:54:21,775 --> 00:54:26,349
and development of software. And
there was a slide with a mission

910
00:54:26,349 --> 00:54:30,852
to democratize access to robots.
And, on the Open Robotics team,

911
00:54:30,852 --> 00:54:35,144
we're like, did we write that?
Or is that their slide? And it

912
00:54:35,144 --> 00:54:39,296
turns out it's their slide. and
so this is where we, have a

913
00:54:39,296 --> 00:54:43,447
really good alignment on what,
Open Robotics mission is and

914
00:54:43,447 --> 00:54:47,599
what Intrinsic is looking to do.
And we're looking for, the

915
00:54:47,599 --> 00:54:51,891
ability to Intrinsic to make
some of these bigger investments

916
00:54:51,891 --> 00:54:55,901
and to do things like the RMW
Zenoh project, which is now

917
00:54:55,901 --> 00:55:00,123
underway. We'd wanted to do that
at Open Robotics for a long

918
00:55:00,123 --> 00:55:04,627
time. We had a proposal, we had
an estimate, but we'd shopped it

919
00:55:04,627 --> 00:55:08,919
around and we're not finding
people who are willing to invest

920
00:55:08,919 --> 00:55:13,000
in making that sort of effort to
improve the core systems.

921
00:55:13,000 --> 00:55:16,855
Audrow Nash: Yeah, for sure. I
was, the ROS boss in, for

922
00:55:16,855 --> 00:55:20,848
Humble. version. And we were
talking about DDS and if that

923
00:55:20,848 --> 00:55:25,116
was possible and how much of a
lift, Zenoh, and if we could be

924
00:55:25,116 --> 00:55:29,109
doing that. And it's just, that
was years before, that was

925
00:55:29,109 --> 00:55:32,000
probably two years before the
acquisition.

926
00:55:32,000 --> 00:55:35,546
Geoff Biggs: Yeah, we, we
actually started looking at

927
00:55:35,546 --> 00:55:39,428
Zenoh for Singapore because in
the hospital DDS was having

928
00:55:39,428 --> 00:55:43,375
problems on their commercial
hide down network, things like

929
00:55:43,375 --> 00:55:47,190
discovery not working across
subnets and so on. And so we

930
00:55:47,190 --> 00:55:50,000
started looking at Zenoh way
back in 2020.

931
00:55:50,000 --> 00:55:54,872
Audrow Nash: Gotcha. So and
then, Tully, would you tell me a

932
00:55:54,872 --> 00:55:59,744
bit more about Intrinsic? Who
are, who is this company? What

933
00:55:59,744 --> 00:56:04,534
are they doing? They wanna
democratize robotics, but how do

934
00:56:04,534 --> 00:56:09,000
they wanna democratize robotics?
these kinds of things.

935
00:56:09,000 --> 00:56:14,008
Tully Foote: Yeah, so Intrinsic,
we're a startup. It's coming out

936
00:56:14,008 --> 00:56:17,860
of the Alphabet, Google X
Moonshot factory. And so

937
00:56:17,860 --> 00:56:22,020
Intrinsic has been going for
several years, incubating

938
00:56:22,020 --> 00:56:26,643
internally. the acquisition was
actually before we announced

939
00:56:26,643 --> 00:56:31,343
what Intrinsic was doing. but
what we're, our current product

940
00:56:31,343 --> 00:56:35,811
that we're talking about is,
automation of industrial work

941
00:56:35,811 --> 00:56:40,665
cells so that you can bring more
of the modern robotics to this

942
00:56:40,665 --> 00:56:45,365
industrial workspace. A lot of
robotics has been very focused

943
00:56:45,365 --> 00:56:49,910
on very simple, repetitive
tasks. And we want to be able to

944
00:56:49,910 --> 00:56:54,225
bring a more modern approach,
including vision tools and

945
00:56:54,225 --> 00:56:58,000
dynamic planning, et cetera, to
this environment.

946
00:56:58,000 --> 00:57:00,000
Geoff Biggs: to

947
00:57:00,000 --> 00:57:04,634
Audrow Nash: And we're doing
that with Flowstate. and flow

948
00:57:04,634 --> 00:57:08,948
state is, it's, it seems
awesome, like from the public

949
00:57:08,948 --> 00:57:13,981
demos, it's like a. Behavior or
like a tree of actions that you

950
00:57:13,981 --> 00:57:18,694
put together and it makes a lot
of the really hard robotics

951
00:57:18,694 --> 00:57:23,727
problems in setting up behavior
for robotic arms and work cells

952
00:57:23,727 --> 00:57:28,521
much easier. And so that's the
intention so you can automate

953
00:57:28,521 --> 00:57:29,000
tasks.

954
00:57:29,000 --> 00:57:33,518
Tully Foote: Exactly, we're
focused on providing the ability

955
00:57:33,518 --> 00:57:37,885
to deploy and, configuration
management and bring a lot of

956
00:57:37,885 --> 00:57:42,101
the technologies from the cloud
deployment space. To the

957
00:57:42,101 --> 00:57:45,941
industrial robotics space so
that you can have your

958
00:57:45,941 --> 00:57:49,706
configuration management, your
deployment control,

959
00:57:49,706 --> 00:57:54,148
introspection, logging, et
cetera, from a web UI instead of

960
00:57:54,148 --> 00:57:58,515
necessarily needing to go in
with a teach pendant to train

961
00:57:58,515 --> 00:58:01,000
each robot to do a specific
task.

962
00:58:01,000 --> 00:58:04,991
Audrow Nash: And then also with
the teach pendant it's if you

963
00:58:04,991 --> 00:58:08,916
have a KUKA arm or an ABB arm or
whatever type of arm It's a

964
00:58:08,916 --> 00:58:13,103
different way of doing it. It's
very like you need someone who's

965
00:58:13,103 --> 00:58:15,000
skilled in that specific make

966
00:58:15,000 --> 00:58:18,949
Tully Foote: exactly. The,
skill, like the perception task

967
00:58:18,949 --> 00:58:22,898
can work with any robot out
there. as long as the robot is

968
00:58:22,898 --> 00:58:26,983
capable of doing what you're
looking for it to do. Likewise,

969
00:58:26,983 --> 00:58:31,000
we can, so we can swap out
sensors, we can swap out robots,

970
00:58:31,000 --> 00:58:38,452
Audrow Nash: and so intrinsic at
the end of What year? So we're

971
00:58:38,452 --> 00:58:45,194
in 2024, at the beginning of
2023, at the end of 2022. so

972
00:58:45,194 --> 00:58:52,527
then Open Robotics, OSRC, the
for profit part of Open Robotics

973
00:58:52,527 --> 00:58:59,860
was bought by Intrinsic, right?
Okay. So Tully, you and I went

974
00:58:59,860 --> 00:59:04,000
over and Geoff, you stayed at
OSRF.

975
00:59:04,000 --> 00:59:08,612
Geoff Biggs: So I was asked to,
so OSRF, effectively changed,

976
00:59:08,612 --> 00:59:13,149
the non profit arm. It had to
change its operation model. So

977
00:59:13,149 --> 00:59:17,761
the board of directors, already
existed, had existed from the

978
00:59:17,761 --> 00:59:22,223
start as a non profit, had to
have a board. That was, Brian

979
00:59:22,223 --> 00:59:26,911
Gerkey, Steve Cousins, and Ryan
Gariepy, three people who have

980
00:59:26,911 --> 00:59:31,674
been involved with, ROS all the
way from the beginning. So they

981
00:59:31,674 --> 00:59:35,682
needed to find a minimal staff
to start operating the

982
00:59:35,682 --> 00:59:40,370
nonprofit. And so they chose,
Vanessa Orsi, who was previously

983
00:59:40,370 --> 00:59:44,906
the CFO at the OSRC. she became
the CEO of Open Robotics, of

984
00:59:44,906 --> 00:59:49,670
OSRF she's very capable. She's
done like 20 nonprofits over her

985
00:59:49,670 --> 00:59:53,904
time, in America. And it's,
really good CEO to have. And

986
00:59:53,904 --> 00:59:58,214
then they asked me to come along
and basically direct the

987
00:59:58,214 --> 01:00:02,826
technical side of things, the
project management and figuring

988
01:00:02,826 --> 01:00:07,136
out what we need to make the
projects work better as open

989
01:00:07,136 --> 01:00:11,673
source projects and so on. and
so That was where basically I

990
01:00:11,673 --> 01:00:15,000
ended up switching to OSRF
rather than OSRF.

991
01:00:15,000 --> 01:00:19,727
Audrow Nash: Let's see with that
transition. So it's just you and

992
01:00:19,727 --> 01:00:23,000
Vanessa that are running OSRF,
at the moment.

993
01:00:23,000 --> 01:00:26,569
Geoff Biggs: Yeah. So we are,
the two of us are the only

994
01:00:26,569 --> 01:00:30,520
staff, full time staff that the
OSRF currently has. We do have

995
01:00:30,520 --> 01:00:34,534
several contract engineers from
a wonderful Argentinian company

996
01:00:34,534 --> 01:00:38,230
called Ekumen that provide
contract engineers who help out

997
01:00:38,230 --> 01:00:41,608
with the, Build, Farm,
Maintenance and Infrastructure

998
01:00:41,608 --> 01:00:45,623
Development and all that sort of
stuff. They're really talented

999
01:00:45,623 --> 01:00:49,000
people, wouldn't survive without
them, I don't think.

1000
01:00:49,000 --> 01:00:51,000
Audrow Nash: They're, great.
they were working with us way

1001
01:00:51,000 --> 01:00:54,056
Geoff Biggs: Oh yeah, we've,
been using them for, yeah,

1002
01:00:54,056 --> 01:00:57,389
they've been working with us
since back in the Willow Garage

1003
01:00:57,389 --> 01:01:00,000
days. They're a wonderful
company, so talented.

1004
01:01:00,000 --> 01:01:01,000
Audrow Nash: huh. So

1005
01:01:01,000 --> 01:01:04,848
Geoff Biggs: and so they are
contract engineers, and they

1006
01:01:04,848 --> 01:01:08,899
work on the infrastructure and
keep things going and develop

1007
01:01:08,899 --> 01:01:13,084
new features and so on. And that
is. The extent of the actual,

1008
01:01:13,084 --> 01:01:16,865
paid staff at the OSRF, and
then, everything else is all

1009
01:01:16,865 --> 01:01:20,848
volunteers, the whole community
coming together. Obviously,

1010
01:01:20,848 --> 01:01:24,696
there's still a big chunk of
that is an intrinsic, former

1011
01:01:24,696 --> 01:01:29,017
Open Robotics engineers from the
OSRC are still heavily involved

1012
01:01:29,017 --> 01:01:33,000
and we're very happy about that
and a long way to continue.

1013
01:01:33,000 --> 01:01:38,254
Audrow Nash: Tully, tell me
about the, so you and I, and

1014
01:01:38,254 --> 01:01:43,977
then pretty much everyone else
at Open Robotics came along to

1015
01:01:43,977 --> 01:01:49,512
Intrinsic. now how, are we
involved still on Open Robotics?

1016
01:01:49,512 --> 01:01:54,953
Where, the ROS 2, Gazebo,
OpenRMF, the infrastructure, how

1017
01:01:54,953 --> 01:02:00,864
are we still involved? And, is
it just like They bought all the

1018
01:02:00,864 --> 01:02:06,493
talent and now, like Intrinsic
bought all the talent and now

1019
01:02:06,493 --> 01:02:12,028
it's just a very small team. Or
how are we supporting, Open

1020
01:02:12,028 --> 01:02:17,000
Robotics to be successful in the
future at Intrinsic?

1021
01:02:17,000 --> 01:02:21,697
Tully Foote: Yeah, so at
Intrinsic, we, strongly believe

1022
01:02:21,697 --> 01:02:26,897
in the value and the mission of,
Open Robotics and the various

1023
01:02:26,897 --> 01:02:31,761
projects we're investing to
support them, supporting teams

1024
01:02:31,761 --> 01:02:35,871
at Intrinsic doing core, ROS,
Gazebo, OpenRMF and

1025
01:02:35,871 --> 01:02:40,987
infrastructure development, as
well as, people in the broader

1026
01:02:40,987 --> 01:02:46,103
team are also contributing back
in some of their overhead, or

1027
01:02:46,103 --> 01:02:50,884
like in their, some of their
small amount of time that is

1028
01:02:50,884 --> 01:02:56,000
available to, on top of their
other primary projects as well.

1029
01:02:56,000 --> 01:02:58,565
Audrow Nash: it's like the
maintenance time. It's actually,

1030
01:02:58,565 --> 01:03:01,043
it's very similar to the
maintenance time we had while we

1031
01:03:01,043 --> 01:03:02,000
were at Open Robotics.

1032
01:03:02,000 --> 01:03:06,113
Tully Foote: Yeah, one of the
analogies is that it's very

1033
01:03:06,113 --> 01:03:10,010
similar to Open Robotics. We
have, effectively one big

1034
01:03:10,010 --> 01:03:14,557
project that we're working for
instead of many little ones. But

1035
01:03:14,557 --> 01:03:18,887
where Intrinsic is values and
investing heavily in all these

1036
01:03:18,887 --> 01:03:23,000
projects and wants to continue
to see them be successful.

1037
01:03:23,000 --> 01:03:27,049
Audrow Nash: Then I'm not. 100
percent sure what can be said,

1038
01:03:27,049 --> 01:03:30,699
but how, like from the
community's perspective, if they

1039
01:03:30,699 --> 01:03:34,748
don't know exactly how Intrinsic
is using various things from

1040
01:03:34,748 --> 01:03:38,796
Open Robotics. They may worry
that Intrinsic cuts off part of

1041
01:03:38,796 --> 01:03:42,580
it and some part of the Open
Robotics project suites dies

1042
01:03:42,580 --> 01:03:46,695
because of lack of investment.
How, did the different parts of

1043
01:03:46,695 --> 01:03:50,876
Open Robotics, and is it fair to
say that, or should I say OSRF

1044
01:03:50,876 --> 01:03:53,000
for this? How, did the different

1045
01:03:53,000 --> 01:03:56,710
Geoff Biggs: Probably OSRF would
be a bit clearer at this point,

1046
01:03:56,710 --> 01:03:57,000
yeah.

1047
01:03:57,000 --> 01:04:02,736
Audrow Nash: So how did the
different parts of OSRF fit in

1048
01:04:02,736 --> 01:04:06,000
Intrinsic within what we can
say,

1049
01:04:06,000 --> 01:04:10,797
Tully Foote: So we're, actively
using all the parts, major

1050
01:04:10,797 --> 01:04:15,925
components of the projects from
OSRF within Intrinsic. I can't

1051
01:04:15,925 --> 01:04:20,556
say much more about exactly how
we're using them, but I,

1052
01:04:20,556 --> 01:04:25,767
reassure you that they're, being
used and look forward to being

1053
01:04:25,767 --> 01:04:28,000
able to talk more about it.

1054
01:04:28,000 --> 01:04:32,000
Tully Foote: But I think that
the, main thing that we want to,

1055
01:04:32,000 --> 01:04:35,935
do, and actually a lot of why
we're here, is actually to make

1056
01:04:35,935 --> 01:04:38,968
it, make sure that the
communities around these

1057
01:04:38,968 --> 01:04:42,710
projects are bigger than one
company. And that's why we've

1058
01:04:42,710 --> 01:04:46,387
actually been pushing toward
what we're just about to get

1059
01:04:46,387 --> 01:04:50,000
talking to about now, the Open
Source Robotics Alliance.

1060
01:04:50,000 --> 01:04:52,000
Audrow Nash: Go ahead.

1061
01:04:52,000 --> 01:04:53,000
Tully Foote: Geoff, you wanna
take this?

1062
01:04:53,000 --> 01:04:57,089
Geoff Biggs: Okay. Yeah. okay.
That was a sudden change. Yeah,

1063
01:04:57,089 --> 01:05:00,651
the Open Source Robotics
Alliance, is a new initiative

1064
01:05:00,651 --> 01:05:04,806
from OSRF and in a nutshell, it
is our plan for how we're going

1065
01:05:04,806 --> 01:05:08,697
to improve. the governance of
all the projects at the OSRF,

1066
01:05:08,697 --> 01:05:12,324
going forward and basically make
the projects much more

1067
01:05:12,324 --> 01:05:15,886
sustainable and long term
viable. Because up till now,

1068
01:05:15,886 --> 01:05:19,579
we've had the Open Source
Robotics Corporation providing

1069
01:05:19,579 --> 01:05:22,745
some funding and staff,
engineers to work on the

1070
01:05:22,745 --> 01:05:26,900
projects. And it was effectively
a benevolent dictator for life

1071
01:05:26,900 --> 01:05:30,923
situation. The community trusted
the Open Robotics, people to

1072
01:05:30,923 --> 01:05:34,880
Keep the projects going and do
right by the community and in

1073
01:05:34,880 --> 01:05:39,167
return, the Open Robotics people
had you know, broad leeway to do

1074
01:05:39,167 --> 01:05:43,322
what they thought was best for
the projects and the funding for

1075
01:05:43,322 --> 01:05:47,279
that as well. But, without the
OSRC anymore, the OSRF has no

1076
01:05:47,279 --> 01:05:50,973
staff and no, no funding to keep
all these things going.

1077
01:05:50,973 --> 01:05:54,798
obviously the Open Source
Robotics Foundation has not gone

1078
01:05:54,798 --> 01:05:59,019
bankrupt all of a sudden. We've
got enough of a nest egg, but we

1079
01:05:59,019 --> 01:06:03,174
do need to have basically money
and people to keep things going

1080
01:06:03,174 --> 01:06:07,263
in the long term. And from our
point of view, we don't want to

1081
01:06:07,263 --> 01:06:11,154
just do the OSRC again. That's,
we'll just be repeating the

1082
01:06:11,154 --> 01:06:15,375
cycle. We want something that's
going to last much longer and be

1083
01:06:15,375 --> 01:06:19,398
much more sustainable. and in
particular, much more community

1084
01:06:19,398 --> 01:06:23,355
driven, community involved than
the OSRC was able to do just

1085
01:06:23,355 --> 01:06:27,247
because of its nature, the
structure of it. And so that for

1086
01:06:27,247 --> 01:06:31,204
that, we have basically, we've
spent about a year and a half

1087
01:06:31,204 --> 01:06:35,029
now putting together the Open
Source Robotics Alliance and

1088
01:06:35,029 --> 01:06:38,920
that is going to hopefully
resolve all the issues we've had

1089
01:06:38,920 --> 01:06:42,943
up till now with governance and
processes and sustainability.

1090
01:06:42,943 --> 01:06:47,164
The way I put it is that we want
to move away from having just a

1091
01:06:47,164 --> 01:06:51,254
few full time engineers at one
company supporting the projects

1092
01:06:51,254 --> 01:06:54,947
to where we have a hundred
contributors, some of many of

1093
01:06:54,947 --> 01:06:58,970
whom may be part time working on
the projects. it's much more

1094
01:06:58,970 --> 01:07:02,993
sustainable to have a hundred
people working half the time on

1095
01:07:02,993 --> 01:07:06,621
the project than to have a
couple of dozen working full

1096
01:07:06,621 --> 01:07:10,446
time on it. Because if one of
the full time people leaves,

1097
01:07:10,446 --> 01:07:14,469
it's a huge loss of time and,
talent. If one of the part time

1098
01:07:14,469 --> 01:07:18,624
people leaves when you've got a
hundred of them, it's much less

1099
01:07:18,624 --> 01:07:22,713
of an impact. And that really is
the way that sustainable open

1100
01:07:22,713 --> 01:07:26,604
source projects work. And so
we're moving more towards that

1101
01:07:26,604 --> 01:07:27,000
model.

1102
01:07:27,000 --> 01:07:31,692
Audrow Nash: Okay. what were
some of the issues with the past

1103
01:07:31,692 --> 01:07:36,154
way of doing things? So you
mentioned the loss between, if

1104
01:07:36,154 --> 01:07:40,538
you have just a few full-time
people, or if you have many

1105
01:07:40,538 --> 01:07:45,154
part-time people, so that's a
clear example of an issue that

1106
01:07:45,154 --> 01:07:49,231
you would wanna fix, in
constructing a new structure.

1107
01:07:49,231 --> 01:07:54,154
But are, there other good, clear
examples of things that you are

1108
01:07:54,154 --> 01:07:57,000
trying to change with this
structure?

1109
01:07:57,000 --> 01:08:00,929
Geoff Biggs: Another thing that
we really want to change is how

1110
01:08:00,929 --> 01:08:04,421
much we have in terms of
resources to do things. So with

1111
01:08:04,421 --> 01:08:07,538
Open Source Robotics
Corporation's funding We were

1112
01:08:07,538 --> 01:08:11,155
able to maintain the projects
and develop them at a slower

1113
01:08:11,155 --> 01:08:14,897
pace than we liked, at least
keep them going forward and pay

1114
01:08:14,897 --> 01:08:18,638
for the infrastructure, which I
learned last year costs more

1115
01:08:18,638 --> 01:08:22,691
than you can possibly imagine to
pay for. And we, we've been able

1116
01:08:22,691 --> 01:08:26,495
to do that, but we want to move
to having a model where we've

1117
01:08:26,495 --> 01:08:30,174
got much more resources to
support these projects. And that

1118
01:08:30,174 --> 01:08:33,791
Basically means financial
resources, we want to be able to

1119
01:08:33,791 --> 01:08:37,720
expand the capabilities of the
infrastructure to have even more

1120
01:08:37,720 --> 01:08:41,149
support for projects or more CI
time, stronger CI time,

1121
01:08:41,149 --> 01:08:45,078
stronger, CI tools like static
analysis to make sure things are

1122
01:08:45,078 --> 01:08:48,632
really production quality,
things like it. Documentation,

1123
01:08:48,632 --> 01:08:52,623
in open source, very few people
want to spend their time writing

1124
01:08:52,623 --> 01:08:56,115
documentation. They want to
write the cool features that

1125
01:08:56,115 --> 01:09:00,044
people are going to use. But the
documentation is still just as

1126
01:09:00,044 --> 01:09:03,785
important. And we, in the past
at Open Robotics, we actually

1127
01:09:03,785 --> 01:09:07,651
had on staff a document writer.
She was very good, but she was

1128
01:09:07,651 --> 01:09:11,019
also just one person. And
there's very much a limit to

1129
01:09:11,019 --> 01:09:14,822
what she was able to achieve
because of that. There's so much

1130
01:09:14,822 --> 01:09:18,315
documentation that needs
written. And so if we have more

1131
01:09:18,315 --> 01:09:21,869
financial resources, the OSRF
would be able to hire more,

1132
01:09:21,869 --> 01:09:25,174
document writers, on an as
needed basis or full time,

1133
01:09:25,174 --> 01:09:27,980
appropriately, in order to
really improve the

1134
01:09:27,980 --> 01:09:31,846
documentation. Another example,
if you want the project leader

1135
01:09:31,846 --> 01:09:34,902
to have The full time
availability to commit to a

1136
01:09:34,902 --> 01:09:38,706
project. We could do that at
Open Source Robotics Corporation

1137
01:09:38,706 --> 01:09:42,010
because they work for us. But
now they work for other

1138
01:09:42,010 --> 01:09:45,191
companies and they'll be
volunteers. if you want to

1139
01:09:45,191 --> 01:09:48,745
guarantee their time, it'd be
nice if you had to pay them

1140
01:09:48,745 --> 01:09:52,424
yourselves. And infrastructure,
not just the infrastructure

1141
01:09:52,424 --> 01:09:56,041
costs, but also the cost of
people to run and maintain the

1142
01:09:56,041 --> 01:09:59,658
infrastructure. developer
advocates, Kat Scott is amazing,

1143
01:09:59,658 --> 01:10:03,212
but again, she's just one
person. We'd love to have, half

1144
01:10:03,212 --> 01:10:06,891
a dozen Kat Scotts just going
around and telling people how

1145
01:10:06,891 --> 01:10:10,383
great our software is and
figuring out what the problems

1146
01:10:10,383 --> 01:10:14,000
are that need to be fixed.
There's always Go ahead, Tully.

1147
01:10:14,000 --> 01:10:18,458
Tully Foote: I want to say I
think one of the things we're,

1148
01:10:18,458 --> 01:10:23,219
looking for with this is to grow
the, this core community, part

1149
01:10:23,219 --> 01:10:27,904
of one of the challenges is that
with the ad hoc nature of the

1150
01:10:27,904 --> 01:10:32,588
governance of the project and
mostly being driven through OSRC

1151
01:10:32,588 --> 01:10:36,895
and the staff there, you really
needed a moderately tight

1152
01:10:36,895 --> 01:10:40,900
connection to connect and be
able to contribute. Like

1153
01:10:40,900 --> 01:10:45,510
obviously, pull requests are
welcome, etc. from outside, but,

1154
01:10:45,510 --> 01:10:50,270
we didn't have a good formalized
method for people to come into

1155
01:10:50,270 --> 01:10:54,653
the project and gain, gain
experience and trust and become

1156
01:10:54,653 --> 01:10:59,111
more and more delegation was
harder to do because we didn't

1157
01:10:59,111 --> 01:11:01,000
have a formalized method.

1158
01:11:01,000 --> 01:11:02,000
Geoff Biggs: Yep.

1159
01:11:01,000 --> 01:11:04,942
Tully Foote: of the things we
did to encourage external

1160
01:11:04,942 --> 01:11:09,457
contribution was we created the
Technical Steering Committee or

1161
01:11:09,457 --> 01:11:13,972
the TSC, and this was explicitly
done. As a way to bring in and

1162
01:11:13,972 --> 01:11:18,559
provide a venue for corporations
to come in and provide guidance

1163
01:11:18,559 --> 01:11:23,002
to the project. we set this up.
Early on, and it's a different

1164
01:11:23,002 --> 01:11:27,231
model than most open source
projects did because we had the

1165
01:11:27,231 --> 01:11:30,957
benefit of the Open Source
Robotics Corporation team

1166
01:11:30,957 --> 01:11:35,114
providing a lot of that guidance
and core capabilities and

1167
01:11:35,114 --> 01:11:39,056
overhead support. we actually
really wanted to make the

1168
01:11:39,056 --> 01:11:43,571
minimum bar threshold for people
to come in and said, if you're

1169
01:11:43,571 --> 01:11:47,226
going to be actively
contributing, we'll give you a

1170
01:11:47,226 --> 01:11:51,240
seat at the table. this is.
Because we had the luxury of

1171
01:11:51,240 --> 01:11:55,683
doing that with the OSRC staff
there and available. You'll see

1172
01:11:55,683 --> 01:11:59,768
a lot of the sort of boilerplate
and background stuff and

1173
01:11:59,768 --> 01:12:04,355
infrastructure was all provided
by OSRC and not the TSC. whereas

1174
01:12:04,355 --> 01:12:05,000
Go ahead.

1175
01:12:05,000 --> 01:12:09,606
Audrow Nash: and this is in the
Open Robotics days when we,

1176
01:12:09,606 --> 01:12:14,368
before the acquisition, we had
the TSC, and yeah, so then you

1177
01:12:14,368 --> 01:12:18,974
had companies have a seat if
they have a full time employee

1178
01:12:18,974 --> 01:12:24,048
on working on ROS or Gazebo, one
of our projects, and then, okay,

1179
01:12:24,048 --> 01:12:26,000
so continuing from there.

1180
01:12:26,000 --> 01:12:30,090
Tully Foote: So I think that
from there, what we're looking

1181
01:12:30,090 --> 01:12:34,181
to do with OSRA is open this up
and make their provide more

1182
01:12:34,181 --> 01:12:37,924
formal channels for the
collaboration across a broader

1183
01:12:37,924 --> 01:12:42,431
community. you mentioned for the
TSC, the one of the requirements

1184
01:12:42,431 --> 01:12:46,105
is to be able to provide a full
time engineer, to the

1185
01:12:46,105 --> 01:12:50,265
development, which is actually a
really high threshold. it's

1186
01:12:50,265 --> 01:12:53,870
basically impossible for an
individual and any small

1187
01:12:53,870 --> 01:12:57,960
startup, that's actually very
hard to do as well. and so we

1188
01:12:57,960 --> 01:12:59,000
wanted to have,

1189
01:12:59,000 --> 01:13:03,227
Audrow Nash: a hundred thousand
dollars of the company. That's

1190
01:13:03,227 --> 01:13:05,000
just going to go right to,

1191
01:13:05,000 --> 01:13:09,219
Tully Foote: exactly, a full
time engineer, you'd be able to

1192
01:13:09,219 --> 01:13:13,578
have a whole staff member and, A
lot of companies that want to

1193
01:13:13,578 --> 01:13:18,008
contribute, maybe they're only a
couple of people, a third or a

1194
01:13:18,008 --> 01:13:21,734
quarter of your workforce to
specifically open source

1195
01:13:21,734 --> 01:13:26,234
development for the platform, is
not feasible at that stage, but

1196
01:13:26,234 --> 01:13:30,594
we want to have paths for those
companies to engage and have a

1197
01:13:30,594 --> 01:13:32,000
voice in the system.

1198
01:13:32,000 --> 01:13:36,455
Audrow Nash: Yeah. So what I'm
hearing, from the previous model

1199
01:13:36,455 --> 01:13:40,768
to the new model, one of the
things that you're really trying

1200
01:13:40,768 --> 01:13:45,010
to do is you're trying to allow
more people to contribute by

1201
01:13:45,010 --> 01:13:49,111
having accepting fractional.
people effectively. So if you

1202
01:13:49,111 --> 01:13:53,283
can contribute a fraction of
your time, you are, you have a

1203
01:13:53,283 --> 01:13:57,808
good structure for how they can
contribute and how they can be a

1204
01:13:57,808 --> 01:14:02,051
part of the community. Whereas
before it was full people and

1205
01:14:02,051 --> 01:14:05,939
that left a lot of, people who
might have been happy to

1206
01:14:05,939 --> 01:14:10,111
contribute, but can't give as
much as was required. So that

1207
01:14:10,111 --> 01:14:14,000
seems to be a big theme in this
OSRA, initiative to me.

1208
01:14:14,000 --> 01:14:17,975
Geoff Biggs: Yeah. We're trying
to move away from tying, being

1209
01:14:17,975 --> 01:14:21,886
involved in the governance as
like a member from how much you

1210
01:14:21,886 --> 01:14:25,284
contribute, because we don't
think that's fair, every

1211
01:14:25,284 --> 01:14:29,388
contribution is very important.
Even if it's just a tiny bug for

1212
01:14:29,388 --> 01:14:33,107
us, it's going to be helping
someone else because, someone

1213
01:14:33,107 --> 01:14:36,953
else is going to have that bug
too. And the more of those we

1214
01:14:36,953 --> 01:14:41,057
have, the better for the project
and the more further forward it

1215
01:14:41,057 --> 01:14:45,032
moves. And so we want to move
away from that. But we also want

1216
01:14:45,032 --> 01:14:48,687
to have a way to get the
financial income that we need to

1217
01:14:48,687 --> 01:14:52,598
support the projects. And so we
are moving away from our, you

1218
01:14:52,598 --> 01:14:56,637
TSC model, with its very loose,
ambiguous processes and No real

1219
01:14:56,637 --> 01:15:00,292
defined way of being more
involved in a project than just

1220
01:15:00,292 --> 01:15:04,267
sending in pull requests. And
we're moving more to a, more of,

1221
01:15:04,267 --> 01:15:08,114
I'm not sure traditional is the
right word since open source

1222
01:15:08,114 --> 01:15:12,025
hasn't been around that long,
but it is more of a traditional

1223
01:15:12,025 --> 01:15:16,000
approach to doing open source
software funding and management.

1224
01:15:16,000 --> 01:15:19,012
Audrow Nash: so is it something
you can look at from like the

1225
01:15:19,012 --> 01:15:20,000
Python community or?

1226
01:15:20,000 --> 01:15:23,637
Geoff Biggs: Yeah, so if you
look at things like the Linux

1227
01:15:23,637 --> 01:15:27,275
Foundation, or the Eclipse
Foundation, or the Cloud Native

1228
01:15:27,275 --> 01:15:31,163
Computing Foundation, there's so
many of these out there these

1229
01:15:31,163 --> 01:15:34,863
days, They all have a very
similar model, where members pay

1230
01:15:34,863 --> 01:15:38,437
a membership fee. on an annual
basis, and that gives them

1231
01:15:38,437 --> 01:15:42,514
certain rights and benefits. For
example, they get the, the right

1232
01:15:42,514 --> 01:15:46,527
to be on some kind of governing
committee as one voice, and they

1233
01:15:46,527 --> 01:15:50,290
get advertising benefits from
being mentioned on the website

1234
01:15:50,290 --> 01:15:54,116
and so on. And so we're going to
have that in the OSRA. There

1235
01:15:54,116 --> 01:15:57,690
will be membership, for,
available for organizations, and

1236
01:15:57,690 --> 01:16:01,265
they can be involved at the high
levels of governance, of

1237
01:16:01,265 --> 01:16:05,278
technical governance of the OSRA
and the OSRA's projects. But we

1238
01:16:05,278 --> 01:16:08,978
also don't want to just say,
okay, if you've got, if you've

1239
01:16:08,978 --> 01:16:12,741
got a million dollars a year
income, give us a chunk of that

1240
01:16:12,741 --> 01:16:16,316
and you can be making the
decisions. that's not very fair

1241
01:16:16,316 --> 01:16:20,267
to the community, which is for
ROS and Gazebo and Open RMAF and

1242
01:16:20,267 --> 01:16:23,716
our infrastructure, the
community is just so important.

1243
01:16:23,716 --> 01:16:27,479
It's the community that has
grown them to this point over 15

1244
01:16:27,479 --> 01:16:31,179
years. And so we want to make
sure that the community still

1245
01:16:31,179 --> 01:16:35,130
has involvement. And so we have
spiced things up a bit from the

1246
01:16:35,130 --> 01:16:39,206
traditional model. we've kept, a
lot of the meritocratic approach

1247
01:16:39,206 --> 01:16:42,655
of some foundations like the
Eclipse Foundation and the

1248
01:16:42,655 --> 01:16:46,167
Apache Foundation for the
project governance. So the day

1249
01:16:46,167 --> 01:16:49,992
to day of the projects, making
the design decisions, deciding

1250
01:16:49,992 --> 01:16:53,881
if a PR should be merged, all
that sort of stuff. It should be

1251
01:16:53,881 --> 01:16:57,581
done by the engineers who are
actually doing it, is what we

1252
01:16:57,581 --> 01:17:01,343
feel. and so the contributors
who, Particularly dedicated to

1253
01:17:01,343 --> 01:17:05,232
the project. they don't have to
be a member of the SRA at all.

1254
01:17:05,232 --> 01:17:08,994
they can apply to be committers
in the project and that gets

1255
01:17:08,994 --> 01:17:12,444
them rights to merge broad quest
and be involved in the

1256
01:17:12,444 --> 01:17:16,081
day-to-day design decisions,
and, all the other stuff that

1257
01:17:16,081 --> 01:17:19,844
goes on every day in a open
source project. and we're having

1258
01:17:19,844 --> 01:17:23,669
a process for that is based on
the Debian process for gaining

1259
01:17:23,669 --> 01:17:27,620
privileges. Where you have, you
apply and you have a mentor who

1260
01:17:27,620 --> 01:17:31,634
mentors you for a certain period
of time and then the People who

1261
01:17:31,634 --> 01:17:35,145
have already gone through that
process and attained that

1262
01:17:35,145 --> 01:17:39,034
committer status They can take a
decision whether you become a

1263
01:17:39,034 --> 01:17:42,796
committer or not. So it's all
entirely meritocratic. there's

1264
01:17:42,796 --> 01:17:46,622
no money involved at all It's
based entirely around Does this

1265
01:17:46,622 --> 01:17:50,385
person contribute? Do they make
good contributions? Are they

1266
01:17:50,385 --> 01:17:54,461
good for the community? Are they
good for the project? And if you

1267
01:17:54,461 --> 01:17:58,224
are, you can be in. It's free.
And that, is the entire lower

1268
01:17:58,224 --> 01:18:01,673
layer of our OSRA governance
structure with our project

1269
01:18:01,673 --> 01:18:05,185
management committees, one for
each project, is entirely

1270
01:18:05,185 --> 01:18:06,000
meritocratic.

1271
01:18:06,000 --> 01:18:09,874
Geoff Biggs: Above that is where
we have more of the traditional

1272
01:18:09,874 --> 01:18:13,566
consortium approach, where we
have what we call the Technical

1273
01:18:13,566 --> 01:18:16,956
Governance Committee. The
Technical Governance Committee

1274
01:18:16,956 --> 01:18:20,709
takes more of a long term view.
they worry about the OSRF as a

1275
01:18:20,709 --> 01:18:24,280
whole, its whole technical
roadmap as a whole, and not just

1276
01:18:24,280 --> 01:18:27,912
not just the ROS roadmap, but
the ROS and Gazebo and OpenRMF

1277
01:18:27,912 --> 01:18:31,604
roadmaps and how do they merge
together into a more of a long

1278
01:18:31,604 --> 01:18:35,115
term vision, and things like
that. And they, that is where

1279
01:18:35,115 --> 01:18:38,626
the consortium members get
their, voice gets heard, and so

1280
01:18:38,626 --> 01:18:42,379
they take the long term view.
But even there, we don't want to

1281
01:18:42,379 --> 01:18:46,192
just say, at the high level, pay
to play, that doesn't work for

1282
01:18:46,192 --> 01:18:49,885
us. And so we split the power.
At the TGC level, we split the

1283
01:18:49,885 --> 01:18:53,819
power between the paying members
and the meritocratic members. So

1284
01:18:53,819 --> 01:18:57,330
representatives of the projects
and representatives of the

1285
01:18:57,330 --> 01:19:00,659
paying members, they come
together and they discuss and

1286
01:19:00,659 --> 01:19:04,412
they make decisions that will
impact the longer term health of

1287
01:19:04,412 --> 01:19:08,226
the foundation and its projects,
for example, how do we use the

1288
01:19:08,226 --> 01:19:12,039
budget to make sure the projects
are getting the resources they

1289
01:19:12,039 --> 01:19:15,852
need, for example, do we need to
hire more document writers? Do

1290
01:19:15,852 --> 01:19:19,000
we need to invest in more CI
time? Things like that.

1291
01:19:19,000 --> 01:19:22,881
Audrow Nash: I like that a lot.
So it's almost like the US

1292
01:19:22,881 --> 01:19:26,829
government or something where
you have multiple branches of

1293
01:19:26,829 --> 01:19:31,178
it. So you have the meritocratic
part and you have the consortium

1294
01:19:31,178 --> 01:19:35,260
based part. And those both make
decisions and they have their

1295
01:19:35,260 --> 01:19:37,000
own weights in some sense.

1296
01:19:37,000 --> 01:19:40,359
Geoff Biggs: Yeah, so we've
tried to balance, we're trying

1297
01:19:40,359 --> 01:19:43,776
to strike a balance basically
between community involvement

1298
01:19:43,776 --> 01:19:47,367
and involvement of people, of
companies who are putting up big

1299
01:19:47,367 --> 01:19:50,436
chunks of money. Basically
because, there's companies

1300
01:19:50,436 --> 01:19:53,969
they're putting up money to help
fund these projects and keep

1301
01:19:53,969 --> 01:19:57,618
them going and they do want to
get something in return and it's

1302
01:19:57,618 --> 01:20:00,977
not just all marketing and
having their logo on a website.

1303
01:20:00,977 --> 01:20:04,568
They want to have, some of them
want to have like engineering,

1304
01:20:04,568 --> 01:20:08,100
voices involved and so that's
why they get to have, some kind

1305
01:20:08,100 --> 01:20:11,517
of voice in the TGC where they
do technical governance as a

1306
01:20:11,517 --> 01:20:15,108
whole and look at how all the
projects are going and make sure

1307
01:20:15,108 --> 01:20:18,815
things are going right and just
you know keep the general health

1308
01:20:18,815 --> 01:20:22,000
of the foundation good for the
benefit of the projects.

1309
01:20:22,000 --> 01:20:23,000
Audrow Nash: I like that quite a
lot.

1310
01:20:23,000 --> 01:20:28,425
Audrow Nash: What, explicitly
are the benefits to joining as a

1311
01:20:28,425 --> 01:20:30,000
consortium member?

1312
01:20:30,000 --> 01:20:32,986
Geoff Biggs: So different
companies look at the benefits

1313
01:20:32,986 --> 01:20:36,131
in different ways like they
depending on what they want and

1314
01:20:36,131 --> 01:20:39,330
what sort of company they are
they have different weights of

1315
01:20:39,330 --> 01:20:42,635
what's important to them. So if
you look at something like the

1316
01:20:42,635 --> 01:20:46,047
Linux Foundation. the companies
that pay huge chunks of money to

1317
01:20:46,047 --> 01:20:49,513
be like the platinum sponsors of
the Vince Foundation, they get a

1318
01:20:49,513 --> 01:20:52,605
board seat, and so they get to
be involved in the decision

1319
01:20:52,605 --> 01:20:55,857
making of the foundation at a
very high level. But that's not

1320
01:20:55,857 --> 01:20:58,896
really what they care about so
much. So long as, the next

1321
01:20:58,896 --> 01:21:02,308
trademark keeps on going and the
project keeps on going, they're

1322
01:21:02,308 --> 01:21:05,666
happy, generally. Because it's
their engineers who are using it

1323
01:21:05,666 --> 01:21:08,919
and product managers who are
making decisions about it and so

1324
01:21:08,919 --> 01:21:12,224
on. But they do the fact that
they can just go around and say,

1325
01:21:12,224 --> 01:21:15,583
Hey, we're a big sponsor of this
and we're, got our logo on the

1326
01:21:15,583 --> 01:21:18,781
website. And they get, primary,
they get priority choice of,

1327
01:21:18,781 --> 01:21:21,500
sponsor booths at Linux
conferences and things like

1328
01:21:21,500 --> 01:21:24,912
that. So for a lot of companies,
it's purely marketing benefits.

1329
01:21:24,912 --> 01:21:28,218
They're just interested in the
various different ways they can

1330
01:21:28,218 --> 01:21:31,470
get marketing benefits out of
it. And that's why you actually

1331
01:21:31,470 --> 01:21:34,722
find that a lot of members of
these sorts of foundations, the

1332
01:21:34,722 --> 01:21:37,601
funding for the membership fee
actually comes from the

1333
01:21:37,601 --> 01:21:40,000
marketing budget. Not the
engineering budget,

1334
01:21:40,000 --> 01:21:41,000
Audrow Nash: Oh, that's funny

1335
01:21:41,000 --> 01:21:43,939
Geoff Biggs: something,
something interesting that I

1336
01:21:43,939 --> 01:21:47,160
learned, it's not quite how I
expected, but it's how they

1337
01:21:47,160 --> 01:21:50,437
work. For other companies, they
want to be involved in the

1338
01:21:50,437 --> 01:21:54,111
technical side of things. And so
that's why they want to be like,

1339
01:21:54,111 --> 01:21:57,558
in our case, they want to be on
the TGC, be involved in those

1340
01:21:57,558 --> 01:22:00,496
technical discussions and
helping to steer the whole

1341
01:22:00,496 --> 01:22:04,000
foundation for the longer term
health of open source robotics.

1342
01:22:04,000 --> 01:22:05,000
Tully Foote: Geoff I can

1343
01:22:04,000 --> 01:22:05,000
Audrow Nash: think, oh, go

1344
01:22:05,000 --> 01:22:08,354
Tully Foote: one other thing,
which is that one of the

1345
01:22:08,354 --> 01:22:11,522
benefits to the companies is
also companies want to

1346
01:22:11,522 --> 01:22:15,000
participate in this to invest in
the project as a whole.

1347
01:22:15,000 --> 01:22:16,000
Geoff Biggs: Yes.

1348
01:22:16,000 --> 01:22:19,718
Tully Foote: if you think about
a company that is building and

1349
01:22:19,718 --> 01:22:22,957
relying on ROS or Gazebo or
something else, if they're

1350
01:22:22,957 --> 01:22:26,615
getting value from leveraging
this open project. and building

1351
01:22:26,615 --> 01:22:30,034
their products and their,
products are building on top of

1352
01:22:30,034 --> 01:22:33,692
this and leveraging this. And if
it was to go away next year,

1353
01:22:33,692 --> 01:22:37,351
that would be very bad for them
because they'd either have to

1354
01:22:37,351 --> 01:22:40,769
purchase another replacement,
build their own, or pick up

1355
01:22:40,769 --> 01:22:43,948
maintenance themselves and so
it's in the interest of

1356
01:22:43,948 --> 01:22:47,846
companies that are out there and
leveraging these tools and these

1357
01:22:47,846 --> 01:22:51,385
resources that are provided in
the open source to make sure

1358
01:22:51,385 --> 01:22:55,043
that the open source project
stays healthy and robust. And so

1359
01:22:55,043 --> 01:22:58,881
this is an opportunity also for
them to provide that support and

1360
01:22:58,881 --> 01:23:02,000
help de risk their own future
looking down the line.

1361
01:23:02,000 --> 01:23:05,036
Geoff Biggs: Yes, you can look
at it as it's like saying

1362
01:23:05,036 --> 01:23:08,181
they're paying their share of
the maintenance costs. Thank

1363
01:23:08,181 --> 01:23:11,217
you. Rather than paying for all
of the maintenance costs

1364
01:23:11,217 --> 01:23:14,524
themselves, and the dozens of
engineers that would take, they

1365
01:23:14,524 --> 01:23:17,669
just pay one small share. And
everyone else pays one small

1366
01:23:17,669 --> 01:23:20,000
share and together everyone
comes together.

1367
01:23:20,000 --> 01:23:23,862
Audrow Nash: That's really nice.
Oh, yeah. And yeah, I've heard

1368
01:23:23,862 --> 01:23:27,724
many companies thinking of it as
Tully said with the de risking

1369
01:23:27,724 --> 01:23:31,648
approach to keeping it going and
one small share for maintenance

1370
01:23:31,648 --> 01:23:35,448
is really nice because then you
can aggregate and you can have

1371
01:23:35,448 --> 01:23:38,881
people who specialize in
maintaining the software rather

1372
01:23:38,881 --> 01:23:42,743
than oh, your team suddenly has
to go figure out some deep down

1373
01:23:42,743 --> 01:23:46,483
bug in ROS 2, which is super
complex. And, someone who's very

1374
01:23:46,483 --> 01:23:50,161
familiar with it on the ROS 2
team would be like, oh yeah, I

1375
01:23:50,161 --> 01:23:52,000
know exactly what's happening.

1376
01:23:52,000 --> 01:23:54,000
Geoff Biggs: Exactly. Yeah.

1377
01:23:54,000 --> 01:23:55,000
Audrow Nash: okay. Hell yeah.

1378
01:23:55,000 --> 01:23:59,136
Audrow Nash: And what, one thing
I believe is that the

1379
01:23:59,136 --> 01:24:03,349
membership, they are tier or so
there, there's probably

1380
01:24:03,349 --> 01:24:08,174
different levels of support, but
also there are different costs

1381
01:24:08,174 --> 01:24:13,000
for different size organizations
and things like this. Correct?

1382
01:24:13,000 --> 01:24:16,793
Geoff Biggs: Yes, so we have
several different types of

1383
01:24:16,793 --> 01:24:19,966
membership. So for
organizations, we have, the

1384
01:24:19,966 --> 01:24:23,897
standard levels is platinum,
gold, and silver, which give

1385
01:24:23,897 --> 01:24:28,172
similar rights, but different
variations of them. And the cost

1386
01:24:28,172 --> 01:24:32,448
for those Also varies based on
the company, company headcount.

1387
01:24:32,448 --> 01:24:36,586
We looked at various different
models, having flat fees and,

1388
01:24:36,586 --> 01:24:40,862
flat fees don't work because if
you're a small company, it's a

1389
01:24:40,862 --> 01:24:45,276
huge amount of money. If you're
Microsoft, it's a tiny amount of

1390
01:24:45,276 --> 01:24:49,414
money and, things like that. so
in the end, we came down to,

1391
01:24:49,414 --> 01:24:53,759
using the, the revenue of either
the company as a whole, or the

1392
01:24:53,759 --> 01:24:58,103
business unit that's involved in
robotics. sorry, income, yeah.

1393
01:24:58,103 --> 01:25:02,310
and that gives, that's a good
way to balance it. because, for

1394
01:25:02,310 --> 01:25:06,586
example, a small start up with
10 people is going to have much

1395
01:25:06,586 --> 01:25:10,655
less income than a very large
company. And At the same time

1396
01:25:10,655 --> 01:25:15,069
also, a very new startup is not
going to have much income at all

1397
01:25:15,069 --> 01:25:19,069
because, they're still just
kicking off, but they may have

1398
01:25:19,069 --> 01:25:21,000
something to contribute even

1399
01:25:21,000 --> 01:25:22,000
Tully Foote: Sorry, Geoff. We do
headcount.

1400
01:25:21,000 --> 01:25:25,846
Geoff Biggs: So that goes a good
Where did we do headcount? Oh,

1401
01:25:25,846 --> 01:25:28,000
sorry, yeah, things changed.

1402
01:25:28,000 --> 01:25:31,382
Audrow Nash: Worries. Yeah. So
it's slightly in flux, but it's

1403
01:25:31,382 --> 01:25:34,000
proportional to the size,
basically on some way.

1404
01:25:34,000 --> 01:25:37,624
Geoff Biggs: It's, the size,
yeah, so it is the size of the,

1405
01:25:37,624 --> 01:25:41,188
company or the business unit
involved because, if it's like

1406
01:25:41,188 --> 01:25:43,000
If you look at something like,

1407
01:25:43,000 --> 01:25:44,000
Audrow Nash: Yeah.

1408
01:25:43,000 --> 01:25:46,000
Geoff Biggs: something like
Microsoft or Bosch or Amazon,

1409
01:25:46,000 --> 01:25:49,000
it's not like the whole thing's
doing robotics, right? So

1410
01:25:49,000 --> 01:25:52,000
Audrow Nash: Yes, for sure.

1411
01:25:52,000 --> 01:25:56,088
Geoff Biggs: Yeah. And so that's
for organizations. We also, of

1412
01:25:56,088 --> 01:25:59,528
course, want academic
institutions and non profits to

1413
01:25:59,528 --> 01:26:03,552
show their support. So we have a
very minimal fee to show that

1414
01:26:03,552 --> 01:26:07,640
support. and most interestingly,
it's something that many other

1415
01:26:07,640 --> 01:26:11,274
foundations don't necessarily
do. We also have a way for

1416
01:26:11,274 --> 01:26:14,000
individuals to be members of the
OSRA. And

1417
01:26:14,000 --> 01:26:16,000
Audrow Nash: At the consortium
level or,

1418
01:26:16,000 --> 01:26:20,107
Geoff Biggs: level, yes. so
individuals can pay a very small

1419
01:26:20,107 --> 01:26:24,213
fee, on an annual basis. And
that gives them the right to be

1420
01:26:24,213 --> 01:26:27,294
what we call end user
representatives. So the

1421
01:26:27,294 --> 01:26:31,263
Technical Governance Committee
and each project management

1422
01:26:31,263 --> 01:26:34,754
committee have end user
representatives on them and

1423
01:26:34,754 --> 01:26:38,861
their job is to represent the
community of users rather than

1424
01:26:38,861 --> 01:26:42,968
developers. So they're looking
at it from a point of view of

1425
01:26:42,968 --> 01:26:47,143
using our software, not just
developing it. And so the people

1426
01:26:47,143 --> 01:26:51,318
who are individual members in
the OSRA are able to be elected

1427
01:26:51,318 --> 01:26:55,219
to these positions and the
elections, the people who vote

1428
01:26:55,219 --> 01:26:59,258
in the elections are the, all
the end user, all the, sorry,

1429
01:26:59,258 --> 01:27:03,296
all the individual members of
the OSRA. Thanks. And so they

1430
01:27:03,296 --> 01:27:07,197
choose one amongst them to
represent them on the TGC, for

1431
01:27:07,197 --> 01:27:11,509
example, and that person's job
is to take the community's voice

1432
01:27:11,509 --> 01:27:15,000
to the TGC, or to the Project
Management Committee.

1433
01:27:15,000 --> 01:27:19,032
Audrow Nash: Okay. That sounds
really cool. And then, All so

1434
01:27:19,032 --> 01:27:22,996
it'll be like, okay, we have a
decision about how we should

1435
01:27:22,996 --> 01:27:27,028
take the direction, like for
this next release, are we going

1436
01:27:27,028 --> 01:27:31,194
to like maybe make a bunch of
features for quad rotors versus,

1437
01:27:31,194 --> 01:27:35,292
and I'm just making stuff up
versus, I don't know, industrial

1438
01:27:35,292 --> 01:27:39,593
robots of some sort. And so then
the community can vote based on

1439
01:27:39,593 --> 01:27:43,692
that, based on their seats in
the consortium, is that kind of

1440
01:27:43,692 --> 01:27:47,253
the intention? And it's
proportional to your level in

1441
01:27:47,253 --> 01:27:49,000
sense, to your head count.

1442
01:27:49,000 --> 01:27:51,986
Geoff Biggs: Yeah, it won't
necessarily be a vote, it

1443
01:27:51,986 --> 01:27:55,254
depends on the decision being
made and how the decision is

1444
01:27:55,254 --> 01:27:58,915
made. if everyone says, hey, I'm
good with that, there's no point

1445
01:27:58,915 --> 01:28:02,577
holding an official vote, right?
Just waste time. But yeah, so if

1446
01:28:02,577 --> 01:28:06,183
it's like a very small, if it's
for an individual project, we're

1447
01:28:06,183 --> 01:28:09,338
making a decision this week
about whether to merge a PR,

1448
01:28:09,338 --> 01:28:12,380
then all people who are
committers on the project plus

1449
01:28:12,380 --> 01:28:15,817
the end user representative on
that project committee will be

1450
01:28:15,817 --> 01:28:19,197
able to, discuss and be involved
in that decision. If it's a

1451
01:28:19,197 --> 01:28:22,521
decision like this project is
asked for, 100k a year to pay

1452
01:28:22,521 --> 01:28:26,127
for a certain resource that they
need to, and they say they need

1453
01:28:26,127 --> 01:28:29,789
it for the project. But if we do
that, we can't give resources to

1454
01:28:29,789 --> 01:28:32,718
these other projects or
something like that, the TGC

1455
01:28:32,718 --> 01:28:35,930
will be involved in making that
decision and the end user

1456
01:28:35,930 --> 01:28:39,366
representative on the TGC will
be part of that discussion and

1457
01:28:39,366 --> 01:28:41,000
part of that decision making.

1458
01:28:41,000 --> 01:28:45,306
Audrow Nash: I see. That's cool.
I like that you're including

1459
01:28:45,306 --> 01:28:47,000
individual contributers.

1460
01:28:47,000 --> 01:28:50,267
Geoff Biggs: Yeah, we think it's
important, the community and the

1461
01:28:50,267 --> 01:28:53,132
individual, they're very
important to us from the history

1462
01:28:53,132 --> 01:28:55,997
of ROS and also robotics,
hobbyists, for example, lots of

1463
01:28:55,997 --> 01:28:58,661
hobbyists in robotics and
there's no reason to ignore

1464
01:28:58,661 --> 01:29:01,778
them. And yeah, we think that
they should be involved and they

1465
01:29:01,778 --> 01:29:04,844
should have a voice, even if
it's just one amongst many, they

1466
01:29:04,844 --> 01:29:06,000
still need to be heard.

1467
01:29:06,000 --> 01:29:10,244
Audrow Nash: Yes, for sure.
okay, that sounds great. So we

1468
01:29:10,244 --> 01:29:14,268
have the consortium based
approach and then we have the

1469
01:29:14,268 --> 01:29:18,000
meritocratic one where you're on
specific projects.

1470
01:29:18,000 --> 01:29:21,546
Audrow Nash: I imagine that
there's a lot of very

1471
01:29:21,546 --> 01:29:26,178
interesting ideas of how to get
this set up and rolling and this

1472
01:29:26,178 --> 01:29:30,592
kind of thing. Maybe, do you
think? Tully, do you think you'd

1473
01:29:30,592 --> 01:29:34,428
be good to speak to this or
should we go to Geoff for

1474
01:29:34,428 --> 01:29:38,987
talking about the meritocratic
part? You go, you guys know what

1475
01:29:38,987 --> 01:29:40,000
I mean anyway.

1476
01:29:40,000 --> 01:29:41,000
Geoff Biggs: Yeah, meritocratic.

1477
01:29:41,000 --> 01:29:47,000
Audrow Nash: Yes. Meritocratic.
go ahead Tully.

1478
01:29:47,000 --> 01:29:51,200
Tully Foote: Yeah, I think what
we'd really like to do is keep,

1479
01:29:51,200 --> 01:29:54,867
we, we have a pretty good
culture of this at the moment

1480
01:29:54,867 --> 01:29:58,667
and we're trying to look to
formalize it. We have project

1481
01:29:58,667 --> 01:30:02,933
leads, we've actually asked all
of the existing project leads to

1482
01:30:02,933 --> 01:30:07,000
continue on in the project lead
role under OSRA, and then for

1483
01:30:07,000 --> 01:30:11,000
the first year and at the end of
the year, they'll be up for

1484
01:30:11,000 --> 01:30:15,067
reelection within the PMC going
forward, in the next year, so

1485
01:30:15,067 --> 01:30:19,333
we'd really like it that we can
basically formalize a little bit

1486
01:30:19,333 --> 01:30:23,133
of what we're doing. Each
project committee will have the

1487
01:30:23,133 --> 01:30:27,267
opportunity to define their own
process. each of our different

1488
01:30:27,267 --> 01:30:31,200
projects has very different
composition and makeup. The ROS

1489
01:30:31,200 --> 01:30:34,800
project has a lot of moderately
federated packages and

1490
01:30:34,800 --> 01:30:38,933
maintainer ship is distributed,
whereas something like, Gazebo

1491
01:30:38,933 --> 01:30:42,533
is midscale with some
federation, but a lot of overlap

1492
01:30:42,533 --> 01:30:46,533
on core maintainers. And then
the OpenRMF and infrastructure

1493
01:30:46,533 --> 01:30:50,800
projects are actually much, much
tighter knit communities with a

1494
01:30:50,800 --> 01:30:55,000
smaller group. And so the, same
project management structure is

1495
01:30:55,000 --> 01:30:59,200
not gonna be correct for each of
them. And so we want to, we're

1496
01:30:59,200 --> 01:31:03,200
gonna let each of those project
committees. Set up their own

1497
01:31:03,200 --> 01:31:07,133
structure. We've laid out some
high level guides saying you

1498
01:31:07,133 --> 01:31:10,867
need to provide structure along
these lines, such as the

1499
01:31:10,867 --> 01:31:15,133
mentorship and adoption process
for new contributors to come in.

1500
01:31:15,133 --> 01:31:19,333
But we're going to leave that up
to each, committee to actually

1501
01:31:19,333 --> 01:31:23,600
define that. And when you become
a committer. What access do you

1502
01:31:23,600 --> 01:31:27,667
get, et cetera. it depends on
what project. It depends on the

1503
01:31:27,667 --> 01:31:30,000
scope and how they want to do
that.

1504
01:31:30,000 --> 01:31:33,711
Audrow Nash: Gotcha. So we are
coming to the end of the time

1505
01:31:33,711 --> 01:31:37,423
that we had booked. Are you guys
running okay with running a

1506
01:31:37,423 --> 01:31:41,258
little bit long, like maybe 15
minutes or something, or do you

1507
01:31:41,258 --> 01:31:42,000
have to run,

1508
01:31:42,000 --> 01:31:43,000
Geoff Biggs: Good.

1509
01:31:42,000 --> 01:31:43,000
Tully Foote: Yeah, I could keep
going.

1510
01:31:43,000 --> 01:31:48,441
Audrow Nash: Okay, hell yeah. I
like the idea of it because of

1511
01:31:48,441 --> 01:31:53,267
these. Different projects,
within OSRA are different in

1512
01:31:53,267 --> 01:31:57,918
nature. You're allowing the
people who have been most

1513
01:31:57,918 --> 01:32:03,447
involved to like the, leaders of
each of them to, decide how to

1514
01:32:03,447 --> 01:32:08,888
roll out these strategies. Will
there be additional, cause one

1515
01:32:08,888 --> 01:32:13,714
thing, like ROS 2, which is
mostly what I've worked on,

1516
01:32:13,714 --> 01:32:18,541
between our projects is
enormous. Like it's, probably a

1517
01:32:18,541 --> 01:32:24,069
hundred repositories across the
whole project and, this is just

1518
01:32:24,069 --> 01:32:26,000
our core repositories.

1519
01:32:26,000 --> 01:32:30,344
Audrow Nash: Where are we
getting the initial mentors

1520
01:32:30,344 --> 01:32:35,426
from? Is it mostly people that
are now at Intrinsic or what do

1521
01:32:35,426 --> 01:32:39,525
we imagine for, this, the
onboarding steps for the

1522
01:32:39,525 --> 01:32:41,000
meritocratic part?

1523
01:32:41,000 --> 01:32:44,779
Geoff Biggs: So I'm actually
working on this at the moment,

1524
01:32:44,779 --> 01:32:48,494
as we're recording, we're
recording this in advance of the

1525
01:32:48,494 --> 01:32:52,402
announcement, but I'm working
with the four project leads and

1526
01:32:52,402 --> 01:32:56,245
we are drawing up the lists of
the initial set of committers

1527
01:32:56,245 --> 01:32:59,960
for each project, and what
specific repositories they need

1528
01:32:59,960 --> 01:33:03,675
to have access to. And that's
going to be our, Our kickoff

1529
01:33:03,675 --> 01:33:07,519
point for each PMC and then from
the operational date of the

1530
01:33:07,519 --> 01:33:11,362
OSRA, from that point forward,
anyone will be able to say, I

1531
01:33:11,362 --> 01:33:14,757
want to be a committer, put
their hand up and then go

1532
01:33:14,757 --> 01:33:18,664
through the mentorship process.
and so we think that probably

1533
01:33:18,664 --> 01:33:22,187
for the first year, At least,
the teams would look very

1534
01:33:22,187 --> 01:33:26,095
similar to how they look now. So
for OpenRMF and Gazebo, it's

1535
01:33:26,095 --> 01:33:30,002
quite internal to Intrinsic. For
ROS, there's a bit of a mix,

1536
01:33:30,002 --> 01:33:33,973
probably about, maybe not half
and half, a fairly good balance

1537
01:33:33,973 --> 01:33:37,688
at the moment of, ex Open
Robotics people at Intrinsic and

1538
01:33:37,688 --> 01:33:41,852
people from other companies such
as Sony. But we think over time,

1539
01:33:41,852 --> 01:33:45,695
as other people become more
involved in the projects, become

1540
01:33:45,695 --> 01:33:49,026
more interested in being
involved in the day to day,

1541
01:33:49,026 --> 01:33:52,869
rather than just sending the
occasional PR, we'll think that

1542
01:33:52,869 --> 01:33:56,713
we'll see more committers from
outside the old Open Robotics

1543
01:33:56,713 --> 01:34:00,428
team start becoming involved.
And that's really what we're

1544
01:34:00,428 --> 01:34:04,143
aiming for, we want to have
these project management codes

1545
01:34:04,143 --> 01:34:08,114
be as diverse as possible and in
all the different meanings of

1546
01:34:08,114 --> 01:34:12,022
diversity. we want all of that
in there so that they are long

1547
01:34:12,022 --> 01:34:15,865
term sustainable, we don't want
to have Bob from the ROS PMC

1548
01:34:15,865 --> 01:34:19,644
left and he's not involved
anymore and we don't know how to

1549
01:34:19,644 --> 01:34:23,488
do half the project. That's not
a situation we want to be in

1550
01:34:23,488 --> 01:34:24,000
anymore.

1551
01:34:24,000 --> 01:34:25,000
Audrow Nash: For sure.

1552
01:34:25,000 --> 01:34:29,033
Audrow Nash: Yeah, one, one
thing that just thinking about

1553
01:34:29,033 --> 01:34:33,135
my conversations on Spaces, on X
and things like this, with

1554
01:34:33,135 --> 01:34:37,654
various people in the community,
especially those that are trying

1555
01:34:37,654 --> 01:34:41,756
to get involved in robotics, I
often recommend them go make

1556
01:34:41,756 --> 01:34:45,650
open source contributions
because of the mentorship that

1557
01:34:45,650 --> 01:34:49,613
you get. Yeah, it's really
amazing because, so you make a

1558
01:34:49,613 --> 01:34:53,715
pull request, the changes you
think you should be adding, a

1559
01:34:53,715 --> 01:34:57,817
new feature, fixing a bug,
whatever it is, and then someone

1560
01:34:57,817 --> 01:35:02,267
who has a lot of experience with
the project, if it's one of the

1561
01:35:02,267 --> 01:35:06,230
intrinsic folks or like one of
the original Open Robotics

1562
01:35:06,230 --> 01:35:10,124
people or anyone in the
community that has been doing it

1563
01:35:10,124 --> 01:35:14,504
for a while. They probably have
great software developer chops,

1564
01:35:14,504 --> 01:35:18,397
and they'll tell you oh, you
could approach this better,

1565
01:35:18,397 --> 01:35:22,291
here's how I would frame it,
here's some formatting, or,

1566
01:35:22,291 --> 01:35:26,393
conventions that I would use
that are good at reducing bugs

1567
01:35:26,393 --> 01:35:30,565
or whatever it might be. and I
feel like that's so valuable,

1568
01:35:30,565 --> 01:35:34,945
like my, time when I started at
Open Robotics, it was just such

1569
01:35:34,945 --> 01:35:38,908
fast learning because of the
feedback of the pull request

1570
01:35:38,908 --> 01:35:43,080
process. and I think people will
become much better software

1571
01:35:43,080 --> 01:35:46,000
engineers from doing this kind
of process.

1572
01:35:46,000 --> 01:35:49,920
Geoff Biggs: I really strongly
agree. I tell younger engineers

1573
01:35:49,920 --> 01:35:53,397
whenever I talk to them about
like why they should open

1574
01:35:53,397 --> 01:35:57,000
source. One of the things I
always say is, The job I'm in

1575
01:35:57,000 --> 01:36:00,730
today and probably most of my
career is not because I got a

1576
01:36:00,730 --> 01:36:04,460
PhD, it's because I did open
source software, since 2003 or

1577
01:36:04,460 --> 01:36:08,443
2004 I've been doing open source
software for robotics and that

1578
01:36:08,443 --> 01:36:12,299
has introduced me to people,
it's given me openings, but more

1579
01:36:12,299 --> 01:36:16,282
importantly it's taught me good
software engineering practices,

1580
01:36:16,282 --> 01:36:19,822
just working with so many
different people, learning how

1581
01:36:19,822 --> 01:36:23,868
they work and how they do things
and figuring out what works for

1582
01:36:23,868 --> 01:36:27,724
them and how it will work for
me, That's given me much better

1583
01:36:27,724 --> 01:36:31,517
skills than I would have got
just staying in one company and

1584
01:36:31,517 --> 01:36:34,994
only doing that company's
processes and just, yeah, and

1585
01:36:34,994 --> 01:36:38,345
talking to other engineers,
even, even in robotics, I

1586
01:36:38,345 --> 01:36:42,138
started at Open Robotics in 2020
and I still learn lots, and

1587
01:36:42,138 --> 01:36:45,994
people like William Woodall,
he's, got incredible C++ skills,

1588
01:36:45,994 --> 01:36:49,787
right? He taught me so much
about C++, and Michael Gray, who

1589
01:36:49,787 --> 01:36:53,897
does OpenRMF again, he taught me
amazing things about C templates

1590
01:36:53,897 --> 01:36:57,943
and now Rust, yep. and so you're
always learning in open source.

1591
01:36:57,943 --> 01:37:01,799
It's great. It's a really good
way just to become and stay in

1592
01:37:01,799 --> 01:37:05,718
an amazing software engineer is
to be involved in open source.

1593
01:37:05,718 --> 01:37:09,006
And I think the most important
thing is never be too

1594
01:37:09,006 --> 01:37:12,799
embarrassed to put in a pull
request. Even if it's one line,

1595
01:37:12,799 --> 01:37:16,466
you put in a pull request,
people will see it, and they'll

1596
01:37:16,466 --> 01:37:20,195
tell you how to make it better,
and you will know next time

1597
01:37:20,195 --> 01:37:24,241
around, and you will learn from
that, and your software will get

1598
01:37:24,241 --> 01:37:25,000
better. And,

1599
01:37:25,000 --> 01:37:28,938
Audrow Nash: And even if you, so
like sometimes I'll be working

1600
01:37:28,938 --> 01:37:32,250
on some side project or
something and I'll be using a

1601
01:37:32,250 --> 01:37:36,000
library and something doesn't
seem how I expect. And so I'll

1602
01:37:36,000 --> 01:37:40,000
make even an issue or sometimes
pull request. and a lot of times

1603
01:37:40,000 --> 01:37:43,875
it points out something that was
either poorly communicated in

1604
01:37:43,875 --> 01:37:47,313
their documentation or
misleading or whatever. And it's

1605
01:37:47,313 --> 01:37:51,250
a good opportunity for them to
improve their project. And a lot

1606
01:37:51,250 --> 01:37:55,063
of times the response that I
get, it's even if I think it's a

1607
01:37:55,063 --> 01:37:58,563
silly question, they're very
open to it and they're very

1608
01:37:58,563 --> 01:37:59,000
helpful

1609
01:37:59,000 --> 01:38:02,162
Geoff Biggs: Yeah, you learn
very quickly, you learn very

1610
01:38:02,162 --> 01:38:05,768
quickly that open source, people
who do open source software, are

1611
01:38:05,768 --> 01:38:09,207
generally open minded, they want
to talk to other people about

1612
01:38:09,207 --> 01:38:12,702
the software, they want to learn
from others about how they can

1613
01:38:12,702 --> 01:38:15,808
make it better, where the
problems are, and they want to

1614
01:38:15,808 --> 01:38:19,081
help other people help their
project. if if someone makes a

1615
01:38:19,081 --> 01:38:22,465
pull request to my project that
fixes, or adds a new feature,

1616
01:38:22,465 --> 01:38:25,960
it's in my interest to help them
brush it up and get it merged,

1617
01:38:25,960 --> 01:38:29,399
because then I don't have to do
it myself, and takes less time

1618
01:38:29,399 --> 01:38:32,783
to help them, than it would to
do it myself, and they benefit

1619
01:38:32,783 --> 01:38:36,000
from that because they learn as
well, we all benefit. It's

1620
01:38:36,000 --> 01:38:38,000
Audrow Nash: And they might get
a feature they wanted.

1621
01:38:38,000 --> 01:38:40,849
Geoff Biggs: Yeah, they get a
feature they wanted, we get a

1622
01:38:40,849 --> 01:38:43,699
new feature in the software,
which is good for other users,

1623
01:38:43,699 --> 01:38:46,597
it grows the software, this is
why the open source software,

1624
01:38:46,597 --> 01:38:49,639
Open Source Robotics Foundation
exists, because this is the way

1625
01:38:49,639 --> 01:38:52,295
we believe. Robotics has grown
and is going to continue

1626
01:38:52,295 --> 01:38:55,000
growing, and that's what we want
to support, this whole,

1627
01:38:55,000 --> 01:38:57,621
Tully Foote: and overall the
development velocity of your

1628
01:38:57,621 --> 01:38:59,000
open source is amazing because

1629
01:38:59,000 --> 01:39:00,000
Geoff Biggs: oh, incredible.

1630
01:39:00,000 --> 01:39:03,147
Tully Foote: You have this large
pool of people, each one

1631
01:39:03,147 --> 01:39:06,736
contributes a little bit and the
whole, everyone moves forward at

1632
01:39:06,736 --> 01:39:09,000
the rate of the sum of the
contributions.

1633
01:39:09,000 --> 01:39:12,966
Geoff Biggs: Yeah, and it never
stops too, right? 24 7 it's

1634
01:39:12,966 --> 01:39:17,000
always getting new stuff coming
in if you've got a resource,

1635
01:39:17,000 --> 01:39:18,000
Audrow Nash: For sure

1636
01:39:17,000 --> 01:39:19,000
Geoff Biggs: always going
forward.

1637
01:39:19,000 --> 01:39:23,509
Audrow Nash: One thought that I
have, and I don't know if you

1638
01:39:23,509 --> 01:39:28,092
guys are planning to do this.
but if not, maybe consider doing

1639
01:39:28,092 --> 01:39:32,379
this. I think that one of the
other big benefits of people

1640
01:39:32,379 --> 01:39:36,297
contributing to open source,
kind of Geoff, as you've

1641
01:39:36,297 --> 01:39:41,028
mentioned, is that you start to
meet people in the community and

1642
01:39:41,028 --> 01:39:45,833
that has given you opportunities
and jobs and this kind of thing.

1643
01:39:45,833 --> 01:39:50,120
What I would think would be a
really cool thing is somehow

1644
01:39:50,120 --> 01:39:54,259
have a leaderboard or it
doesn't, need to be competitive

1645
01:39:54,259 --> 01:39:58,103
per se, but some way of making
it so that people are

1646
01:39:58,103 --> 01:40:01,725
highlighted for their
contributions, because that

1647
01:40:01,725 --> 01:40:06,308
visibility will make it so other
people know them a little bit

1648
01:40:06,308 --> 01:40:10,522
better. And maybe it's a better
way to get a first job or

1649
01:40:10,522 --> 01:40:12,000
something like this.

1650
01:40:12,000 --> 01:40:13,000
Geoff Biggs: Oh, yeah,

1651
01:40:13,000 --> 01:40:14,000
Tully Foote: we,

1652
01:40:14,000 --> 01:40:16,000
Audrow Nash: would love to see
that. Go ahead, Tully.

1653
01:40:16,000 --> 01:40:19,368
Tully Foote: I was going to say
we're, actively working on that.

1654
01:40:19,368 --> 01:40:22,368
We're working on a badging
system. So that people can get

1655
01:40:22,368 --> 01:40:25,316
acknowledgement for being
whatever their status is. They

1656
01:40:25,316 --> 01:40:28,684
can show that off in advertising
it. We're also going to work to

1657
01:40:28,684 --> 01:40:31,947
try and integrate it with some
of our various infrastructures.

1658
01:40:31,947 --> 01:40:35,053
See if we can get them badging
on some of our sites and the

1659
01:40:35,053 --> 01:40:36,000
forums, et cetera.

1660
01:40:36,000 --> 01:40:38,810
Audrow Nash: Hell yeah. Oh, I
love it. Go ahead, Geoff. Did

1661
01:40:38,810 --> 01:40:40,000
you have anything to add?

1662
01:40:40,000 --> 01:40:43,298
Geoff Biggs: Yeah, I was just
going to say, apart from the

1663
01:40:43,298 --> 01:40:46,482
badge system, this is actually
something that we've been

1664
01:40:46,482 --> 01:40:49,836
promoting to potential member
companies as well, is that if

1665
01:40:49,836 --> 01:40:53,247
you get involved in the OSRA,
it's a really good way to find

1666
01:40:53,247 --> 01:40:56,431
new talent that you can hire.
Because the people who are

1667
01:40:56,431 --> 01:40:59,104
actively involved at the
meritocratic layer are

1668
01:40:59,104 --> 01:41:02,288
guaranteed to be people who
already know the software so

1669
01:41:02,288 --> 01:41:05,870
they can hit the ground running
when you hire them. But they're

1670
01:41:05,870 --> 01:41:09,054
also guaranteed to be good
engineers who are open minded

1671
01:41:09,054 --> 01:41:12,408
and willing to learn and are
already, already learning just

1672
01:41:12,408 --> 01:41:15,649
by being involved. And so
that's, it's good for companies

1673
01:41:15,649 --> 01:41:19,003
and it's good for the people who
are contributing. everyone

1674
01:41:19,003 --> 01:41:22,585
learns, everyone benefits, and
everyone gets in touch with each

1675
01:41:22,585 --> 01:41:26,110
other. You can see, I know who
that person is, I've seen their

1676
01:41:26,110 --> 01:41:29,579
contributions going into the
software, I think they're great,

1677
01:41:29,579 --> 01:41:31,000
I want to work with them.

1678
01:41:31,000 --> 01:41:36,195
Audrow Nash: Hell yeah. Do you
ever think, so related to that,

1679
01:41:36,195 --> 01:41:41,222
it seems like a fairly logical
next step to me to also help,

1680
01:41:41,222 --> 01:41:45,914
almost have a jobs board. Or for
some of the, consortium

1681
01:41:45,914 --> 01:41:50,689
companies that may be looking
for jobs, like exposing and

1682
01:41:50,689 --> 01:41:55,968
helping people that are looking
connect to possible openings in

1683
01:41:55,968 --> 01:42:01,162
jobs, any plan to do that, or
that, to be, that could be quite

1684
01:42:01,162 --> 01:42:02,000
lucrative,

1685
01:42:02,000 --> 01:42:05,536
Geoff Biggs: That's actually one
of the, one of the membership

1686
01:42:05,536 --> 01:42:09,244
benefits is that, we haven't got
it set up yet, but there will be

1687
01:42:09,244 --> 01:42:12,837
a jobs advertisement system on
the website, and members will be

1688
01:42:12,837 --> 01:42:16,317
able to advertise jobs there.
and, we've seen some of this in

1689
01:42:16,317 --> 01:42:19,340
this in discourse, and, we've
seen people advertising

1690
01:42:19,340 --> 01:42:22,591
internships and full time
positions and so on. We want to

1691
01:42:22,591 --> 01:42:26,184
basically try and formalize it
and make it better for people to

1692
01:42:26,184 --> 01:42:29,607
use. And so for members, this
will be a benefit is that they

1693
01:42:29,607 --> 01:42:33,086
can send in their jobs, that job
board, and then that will be

1694
01:42:33,086 --> 01:42:36,508
somewhere where one central
place for engineers who know ROS

1695
01:42:36,508 --> 01:42:40,045
or know Gazebo can go and find
jobs that are going to be using

1696
01:42:40,045 --> 01:42:43,581
the skills they have and know
that they will have an advantage

1697
01:42:43,581 --> 01:42:46,604
in applying for those jobs
because they'll be able to

1698
01:42:46,604 --> 01:42:49,000
promote their own involvement in
the OSRA.

1699
01:42:49,000 --> 01:42:52,000
Audrow Nash: I really like that.

1700
01:42:52,000 --> 01:42:56,126
Audrow Nash: Okay, now we're
talking about getting people in

1701
01:42:56,126 --> 01:43:00,527
jobs and what this makes me then
think of is skilling people up,

1702
01:43:00,527 --> 01:43:04,308
educating them this kind of
thing so that they are in a

1703
01:43:04,308 --> 01:43:08,090
position to do this. And part of
that is the mentorship

1704
01:43:08,090 --> 01:43:12,079
perspective, but is there going
to be other efforts around

1705
01:43:12,079 --> 01:43:16,273
education? Maybe so that we can
increase the number of people

1706
01:43:16,273 --> 01:43:20,536
that would be contributing. You
mentioned Better Docs as a big

1707
01:43:20,536 --> 01:43:24,043
initiative. but is there
anything else that will go

1708
01:43:24,043 --> 01:43:27,000
towards skilling up people in
general? It's

1709
01:43:27,000 --> 01:43:30,652
Geoff Biggs: We, the Open OSRF
actually has, as one of its

1710
01:43:30,652 --> 01:43:34,618
goals, is education in robotics,
open source software robotics,

1711
01:43:34,618 --> 01:43:38,521
and so we do actually have some
plans in this area, so I think

1712
01:43:38,521 --> 01:43:42,110
everyone knows, the TurtleBot
has been around for a dozen

1713
01:43:42,110 --> 01:43:46,013
years now, I think, started at
World of Garage first, I think,

1714
01:43:46,013 --> 01:43:49,728
yeah, Tully was the originator
of that. As I recall. And so

1715
01:43:49,728 --> 01:43:53,064
that's a big thing for
education, but we also want to

1716
01:43:53,064 --> 01:43:57,031
do, better in that regard. Not
just providing a robot that runs

1717
01:43:57,031 --> 01:44:00,745
ROS, but we actually want to
have, tutorials that work with

1718
01:44:00,745 --> 01:44:04,775
it, and courses you can do with
it, and so on. And so we do have

1719
01:44:04,775 --> 01:44:08,300
ideas in there, but these are
going to take a little bit

1720
01:44:08,300 --> 01:44:10,000
longer to come to fruition.

1721
01:44:10,000 --> 01:44:12,000
Audrow Nash: It's a lot that's
already happening.

1722
01:44:12,000 --> 01:44:14,660
Geoff Biggs: We've focusing on
OSRA for the first year, and,

1723
01:44:14,660 --> 01:44:17,542
once that's up and running, then
we'll have more time to focus on

1724
01:44:17,542 --> 01:44:20,246
those as well. But yeah, watch
the space. There's going to be

1725
01:44:20,246 --> 01:44:21,000
fun stuff coming.

1726
01:44:21,000 --> 01:44:25,870
Audrow Nash: Hell yeah. That
seems awesome to me. is there

1727
01:44:25,870 --> 01:44:30,573
anything else for either of you
that you want to mention

1728
01:44:30,573 --> 01:44:35,947
regarding OSRA? Or I don't know
any, and if not, I would love to

1729
01:44:35,947 --> 01:44:41,237
see where you think things are
going, for the community and for

1730
01:44:41,237 --> 01:44:43,000
our whole initiative.

1731
01:44:43,000 --> 01:44:46,215
Geoff Biggs: I think, for the
community as a whole, not just

1732
01:44:46,215 --> 01:44:49,323
the initiative. obviously, I
hope the OSRA succeeds, and I

1733
01:44:49,323 --> 01:44:52,538
think it will. I think for the
community as a whole, this is

1734
01:44:52,538 --> 01:44:55,753
going to be really a beneficial
thing, very much in the long

1735
01:44:55,753 --> 01:44:58,968
term, especially. in the short
term, if you're a user of the

1736
01:44:58,968 --> 01:45:02,076
software, you're not going to
see much change. If you're a

1737
01:45:02,076 --> 01:45:05,183
developer of the software who
just sends an occasional PR,

1738
01:45:05,183 --> 01:45:08,505
then you're probably not going
to see much change for a while.

1739
01:45:08,505 --> 01:45:11,828
If she's really getting heavily
involved, then maybe they will

1740
01:45:11,828 --> 01:45:14,882
be changed. you could be a
committer, for example, but in

1741
01:45:14,882 --> 01:45:18,097
the longterm, I think this is
really going to accelerate the

1742
01:45:18,097 --> 01:45:21,312
development of all of the OSRF
projects and make them better

1743
01:45:21,312 --> 01:45:24,580
for the people who want to use
them. So if you're a hobbyist,

1744
01:45:24,580 --> 01:45:27,956
it's going to be higher quality
software. So it'd be less. Less

1745
01:45:27,956 --> 01:45:31,332
gotchas for you to worry about,
or less problems. But if you're

1746
01:45:31,332 --> 01:45:34,547
a company, there's going to be
much more production quality.

1747
01:45:34,547 --> 01:45:37,762
There'll be much less, trouble
for you to ensure your robots

1748
01:45:37,762 --> 01:45:41,245
work. Because you will know that
the underlying software is being

1749
01:45:41,245 --> 01:45:44,513
developed using the resources of
the whole robotics industry,

1750
01:45:44,513 --> 01:45:47,728
pulled together to provide great
software. The same way that

1751
01:45:47,728 --> 01:45:50,890
companies can rely on the Linux
kernel for putting in their

1752
01:45:50,890 --> 01:45:54,158
products, I think, really, we're
going to see acceleration of

1753
01:45:54,158 --> 01:45:57,266
that to that level, and also
acceleration of new features.

1754
01:45:57,266 --> 01:46:00,320
just in the past year, the RMW
Zenith thing is just shows

1755
01:46:00,320 --> 01:46:03,642
what's possible when we have a
consortium approach rather than

1756
01:46:03,642 --> 01:46:06,000
just one company doing the full
development.

1757
01:46:06,000 --> 01:46:10,538
Audrow Nash: That's a good
point. Yeah. We are able to make

1758
01:46:10,538 --> 01:46:13,000
a heavy lift now under Intrinsic

1759
01:46:13,000 --> 01:46:14,000
Geoff Biggs: Yeah,

1760
01:46:13,000 --> 01:46:16,587
Audrow Nash: Or Intrinsic
funding a lot of development.

1761
01:46:16,587 --> 01:46:20,761
that's a wonderful thing. Yeah.
And so you're saying more of the

1762
01:46:20,761 --> 01:46:22,000
same will continue.

1763
01:46:22,000 --> 01:46:23,000
Geoff Biggs: I believe so, yes.

1764
01:46:23,000 --> 01:46:26,602
Audrow Nash: And we'll be able
to make larger shifts that

1765
01:46:26,602 --> 01:46:29,825
improve, and it might be
documentation. It might be

1766
01:46:29,825 --> 01:46:33,364
education. It might be, I don't
know, all sorts of other

1767
01:46:33,364 --> 01:46:37,030
supporting things that make the
community richer. Yeah. So

1768
01:46:37,030 --> 01:46:40,000
that's all wonderful. Tully,
what do you think?

1769
01:46:40,000 --> 01:46:43,709
Tully Foote: Yeah, I'm really
want to look to grow and

1770
01:46:43,709 --> 01:46:47,966
diversify. we've been a small
tight knit community and I think

1771
01:46:47,966 --> 01:46:52,087
we, I'd like to see more people
get involved and doing this.

1772
01:46:52,087 --> 01:46:56,139
Setting up the slightly more
formal structure, I think will

1773
01:46:56,139 --> 01:47:00,260
give us the ability to reach out
and grow effectively. We've

1774
01:47:00,260 --> 01:47:03,968
reached the limit of the scale
we can grow with direct

1775
01:47:03,968 --> 01:47:07,883
connections and where I'm
looking forward to. Seeing more

1776
01:47:07,883 --> 01:47:11,935
people getting involved, more
companies getting involved at

1777
01:47:11,935 --> 01:47:16,124
more levels with these, more
opportunities for engagement and

1778
01:47:16,124 --> 01:47:20,313
representation, actually, to
get, more voices involved in the

1779
01:47:20,313 --> 01:47:21,000
community.

1780
01:47:21,000 --> 01:47:26,000
Audrow Nash: Hell yeah. Love it.
Let's see.

1781
01:47:26,000 --> 01:47:31,166
Audrow Nash: Now, starting to
wrap up, what do you guys think

1782
01:47:31,166 --> 01:47:36,078
for robotics, like, when you
think about five years in the

1783
01:47:36,078 --> 01:47:41,160
future, what are you thinking
will be different, where, just

1784
01:47:41,160 --> 01:47:46,580
tell me where you think robotics
will be in five years, and what

1785
01:47:46,580 --> 01:47:51,492
opportunities may be on the way
to there. Wanna start with

1786
01:47:51,492 --> 01:47:52,000
Geoff?

1787
01:47:52,000 --> 01:47:56,113
Geoff Biggs: Okay, I think we're
probably going to see two major

1788
01:47:56,113 --> 01:47:59,969
trends. One is we're going to
see a lot more multiple robots

1789
01:47:59,969 --> 01:48:03,889
in a single space. And I don't
just mean like a whole pile of

1790
01:48:03,889 --> 01:48:07,616
Kivas. multiple robots from
multiple vendors. So like that

1791
01:48:07,616 --> 01:48:11,407
whole thing with, hotels with,
delivery robots going to the

1792
01:48:11,407 --> 01:48:15,327
rooms and cleaning robots and so
on. that sort of thing we're

1793
01:48:15,327 --> 01:48:18,990
going to see much more of, I
think. all these, single use

1794
01:48:18,990 --> 01:48:22,717
service robots are really going
to be everywhere, I think,

1795
01:48:22,717 --> 01:48:26,830
inside and outside. And I think
we're going to see a lot more of

1796
01:48:26,830 --> 01:48:30,429
that. The other thing I think
we're going to see is that

1797
01:48:30,429 --> 01:48:34,285
robots start coming to market a
lot faster. And that will in

1798
01:48:34,285 --> 01:48:38,269
part be because of software like
ROS and in part be because of

1799
01:48:38,269 --> 01:48:41,996
software built on ROS will start
becoming much more widely

1800
01:48:41,996 --> 01:48:46,045
available both open source and
for sale stuff so like you'll be

1801
01:48:46,045 --> 01:48:49,965
able to, we already see a few
companies doing this but you'll

1802
01:48:49,965 --> 01:48:53,370
be able to go and buy a robot
chassis and then from a

1803
01:48:53,370 --> 01:48:57,290
different company go and buy a
navigation system that you can

1804
01:48:57,290 --> 01:49:01,082
put on it for example and so
it'll become much more of that

1805
01:49:01,082 --> 01:49:04,681
robotics companies become
companies who provide services

1806
01:49:04,681 --> 01:49:08,536
using robots, and they just, do
a bit of integration work to

1807
01:49:08,536 --> 01:49:12,199
pull this together. I think
that's the way we're going to

1808
01:49:12,199 --> 01:49:16,248
see things going. And, I really
look forward to getting to that

1809
01:49:16,248 --> 01:49:20,232
point, because that's when the
promise of ROS, and Gazebo, and

1810
01:49:20,232 --> 01:49:24,088
OpenRMF, and open source robots
in general really takes off.

1811
01:49:24,088 --> 01:49:28,072
When people start not building
the whole robot themselves, but

1812
01:49:28,072 --> 01:49:30,000
trading parts with each other.

1813
01:49:30,000 --> 01:49:33,632
Audrow Nash: Yeah, I think I had
an interview, I think the last

1814
01:49:33,632 --> 01:49:37,263
one published, or episode three
was with Polymath Robotics, and

1815
01:49:37,263 --> 01:49:40,549
they're doing just this. And
it's the start of this trend

1816
01:49:40,549 --> 01:49:43,662
from my perspective, where
they're taking ROS, they're

1817
01:49:43,662 --> 01:49:47,351
building some nice things on top
of it that make it very easy to

1818
01:49:47,351 --> 01:49:50,637
access, and then you can just
build applications on that,

1819
01:49:50,637 --> 01:49:53,000
which is so cool. It's starting.
I think.

1820
01:49:53,000 --> 01:49:54,000
Geoff Biggs: It's starting.

1821
01:49:53,000 --> 01:49:55,455
Audrow Nash: Your observation is
already starting, which is the

1822
01:49:55,455 --> 01:49:56,000
coolest thing.

1823
01:49:56,000 --> 01:50:00,706
Tully Foote: Yeah, it's, it's
fun to reflect on that because

1824
01:50:00,706 --> 01:50:05,098
actually one of the visions that
we set out when we were

1825
01:50:05,098 --> 01:50:09,647
designing ROS back in 07, 08
era. Was, we wanted to be the

1826
01:50:09,647 --> 01:50:12,000
LAMP stamp for robotics, where

1827
01:50:12,000 --> 01:50:14,000
Audrow Nash: Does LAMP stamp
mean? I don't know what that,

1828
01:50:14,000 --> 01:50:18,485
Tully Foote: LAMP stack. The
Linux, Apache, MySQL, PHP, Perl,

1829
01:50:18,485 --> 01:50:19,000
Python.

1830
01:50:19,000 --> 01:50:21,000
Audrow Nash: It was how people
would build websites,

1831
01:50:21,000 --> 01:50:24,898
Tully Foote: Exactly. This was
one of the really early

1832
01:50:24,898 --> 01:50:29,086
demonstrations of the power of
open source with those open

1833
01:50:29,086 --> 01:50:33,778
source tools as generic building
blocks. If you had an idea for a

1834
01:50:33,778 --> 01:50:38,326
website that you wanted to stand
up. You could use those tools,

1835
01:50:38,326 --> 01:50:42,008
and a little bit of
configuration and coding on top

1836
01:50:42,008 --> 01:50:46,123
of it, create a website very
quickly. Literally you could

1837
01:50:46,123 --> 01:50:48,000
make prototypes overnight.

1838
01:50:48,000 --> 01:50:51,211
Geoff Biggs: Yeah, you can look
at it as like with that, with

1839
01:50:51,211 --> 01:50:54,474
those four tools you could build
Amazon. That's the way it was

1840
01:50:54,474 --> 01:50:55,000
basically.

1841
01:50:55,000 --> 01:50:56,000
Tully Foote: Yeah,

1842
01:50:56,000 --> 01:50:57,000
Audrow Nash: Yeah.

1843
01:50:57,000 --> 01:51:00,838
Tully Foote: And Amazon and
Facebook, the, all of, if you

1844
01:51:00,838 --> 01:51:05,147
had an idea, and the nice thing
is with these open source tools,

1845
01:51:05,147 --> 01:51:08,918
they were accessible to an
individual. So that if I came

1846
01:51:08,918 --> 01:51:12,890
along, I had a great idea, I
could make a prototype, and it

1847
01:51:12,890 --> 01:51:17,267
would just be there and be ready
to go. And the vision we had for

1848
01:51:17,267 --> 01:51:21,307
ROS was to provide that LAMP
stack for robotics, so that all

1849
01:51:21,307 --> 01:51:25,683
the tools are there that you can
bring together. And, if you have

1850
01:51:25,683 --> 01:51:29,791
an idea for a robot application,
you pick these parts off the

1851
01:51:29,791 --> 01:51:33,628
shelf, the open source ones, you
put them together in the

1852
01:51:33,628 --> 01:51:37,736
configuration that will achieve
your application, and you can

1853
01:51:37,736 --> 01:51:41,776
test it out and deploy very
quickly. And I think we are now,

1854
01:51:41,776 --> 01:51:45,815
we're, that is insight where we
can actually have this thing

1855
01:51:45,815 --> 01:51:49,384
where it may actually be like
robots that are getting

1856
01:51:49,384 --> 01:51:51,000
deployed. It's not just,

1857
01:51:51,000 --> 01:51:52,000
Audrow Nash: Websites

1858
01:51:52,000 --> 01:51:54,000
Tully Foote: Exactly

1859
01:51:54,000 --> 01:51:57,010
Geoff Biggs: We're right at the
very, yeah, we're at the very

1860
01:51:57,010 --> 01:52:00,020
edge of that. And that's, going
to be the next revolution for

1861
01:52:00,020 --> 01:52:02,980
us. I think we've seen this
sudden surge in robotic startups

1862
01:52:02,980 --> 01:52:06,039
for the next one is going to be
when they start sharing stuff,

1863
01:52:06,039 --> 01:52:09,000
selling stuff to each other, and
that's going to be amazing.

1864
01:52:09,000 --> 01:52:10,000
Audrow Nash: Hell yeah.

1865
01:52:10,000 --> 01:52:14,149
Audrow Nash: To maybe ask an
obvious question that we all

1866
01:52:14,149 --> 01:52:18,153
agree with, but may not be
clear, and Tully, what's the

1867
01:52:18,153 --> 01:52:22,594
value of the open source part of
this versus closed source or

1868
01:52:22,594 --> 01:52:27,180
like things you can just buy and
you don't know what's going on

1869
01:52:27,180 --> 01:52:29,000
in them? Why open source?

1870
01:52:29,000 --> 01:52:33,120
Tully Foote: So there, there are
lots of components to open

1871
01:52:33,120 --> 01:52:37,031
source that provide value. I
think that for me, the most

1872
01:52:37,031 --> 01:52:40,174
valuable thing is the
development velocity. I

1873
01:52:40,174 --> 01:52:43,945
mentioned that a little bit
earlier, but basically. If

1874
01:52:43,945 --> 01:52:48,136
there's 10 of us putting little
incremental changes into one

1875
01:52:48,136 --> 01:52:52,535
product for a tenth of our time,
that's actually the equivalent

1876
01:52:52,535 --> 01:52:57,005
of a full time engineer. And so
if you can get this community of

1877
01:52:57,005 --> 01:53:00,846
small contributions that keep
stepping state of the art

1878
01:53:00,846 --> 01:53:05,106
forward, you can actually do
more than if you invested a full

1879
01:53:05,106 --> 01:53:09,297
time person to develop it for
yourself. And so by leveraging

1880
01:53:09,297 --> 01:53:12,998
this broader community, the
development velocity goes

1881
01:53:12,998 --> 01:53:16,979
forward faster. in addition, and
every incremental person

1882
01:53:16,979 --> 01:53:21,099
contributing and participating
in the community accelerates

1883
01:53:21,099 --> 01:53:25,220
everybody together, which is
this really powerful incentive

1884
01:53:25,220 --> 01:53:29,759
to join and collaborate. Because
if you break off and do your own

1885
01:53:29,759 --> 01:53:33,670
little thing, you do that, but
then you incur all of the

1886
01:53:33,670 --> 01:53:38,000
development costs and all of the
overhead and maintenance and,

1887
01:53:38,000 --> 01:53:39,000
Audrow Nash: the tech debt.

1888
01:53:39,000 --> 01:53:43,415
Tully Foote: have to try to keep
up. With the distributed

1889
01:53:43,415 --> 01:53:47,754
community development. so you
have to both invest in the

1890
01:53:47,754 --> 01:53:50,000
velocity and the maintenance.

1891
01:53:50,000 --> 01:53:51,000
Audrow Nash: Go ahead, Geoff.

1892
01:53:51,000 --> 01:53:54,465
Geoff Biggs: The way I like to
put it is that, for the cost of

1893
01:53:54,465 --> 01:53:57,929
a little bit of engineering time
to contribute to the project,

1894
01:53:57,929 --> 01:54:01,282
you get a very large amount of
free labor, basically. That's

1895
01:54:01,282 --> 01:54:04,524
the way I like to put it,
because to companies, free labor

1896
01:54:04,524 --> 01:54:08,044
sounds great compared to paying
for their own engineers, right?

1897
01:54:08,044 --> 01:54:11,565
And that's what you get. you're
getting all these other people.

1898
01:54:11,565 --> 01:54:15,085
are effectively working for you
for free. sure, they're working

1899
01:54:15,085 --> 01:54:18,271
for themselves and for everyone
else as well, but they're

1900
01:54:18,271 --> 01:54:21,568
working for you for free for
something that you could never

1901
01:54:21,568 --> 01:54:24,753
do yourself because it just
takes too much time to do and

1902
01:54:24,753 --> 01:54:28,106
you'd have to pay dozens of
engineers or more. it's really a

1903
01:54:28,106 --> 01:54:29,000
massive benefit.

1904
01:54:29,000 --> 01:54:33,116
Audrow Nash: Yeah. And also like
the hardening of the code too,

1905
01:54:33,116 --> 01:54:37,035
because you have so many people
using it and you're going to

1906
01:54:37,035 --> 01:54:41,216
find a lot more of those flaky 1
percent or one in a thousand or

1907
01:54:41,216 --> 01:54:42,000
10, 000 bugs

1908
01:54:42,000 --> 01:54:46,094
Geoff Biggs: That's actually a
key point there. It's not just

1909
01:54:46,094 --> 01:54:49,852
the number of people, it's the
variety of viewpoints and

1910
01:54:49,852 --> 01:54:52,000
thought processes and use cases.

1911
01:54:52,000 --> 01:54:55,000
Audrow Nash: Great way to put
it. Yeah, Tully.

1912
01:54:55,000 --> 01:54:58,831
Tully Foote: I was going say the
ability to understand that this

1913
01:54:58,831 --> 01:55:02,303
has been used and demonstrated
by a lot of people in a lot

1914
01:55:02,303 --> 01:55:06,014
places. Allows you to have more
confidence in something like a

1915
01:55:06,014 --> 01:55:09,725
heavily used project, and, the
more people have tested it out,

1916
01:55:09,725 --> 01:55:13,377
the more corner cases you've
discovered, the longer it's been

1917
01:55:13,377 --> 01:55:17,028
there, opportunities to find
those defects come up, and if we

1918
01:55:17,028 --> 01:55:20,560
have good processes and good
quality processes to make sure

1919
01:55:20,560 --> 01:55:23,852
that we add regression tests,
For each corner case that

1920
01:55:23,852 --> 01:55:27,504
anybody discovers, you can be
more confident when you pick up

1921
01:55:27,504 --> 01:55:29,000
the open source software.

1922
01:55:29,000 --> 01:55:33,421
Audrow Nash: Definitely. Hell
yeah, okay, that's really cool

1923
01:55:33,421 --> 01:55:36,000
to hear all those thoughts on
this.

1924
01:55:36,000 --> 01:55:41,253
Audrow Nash: If you, were to
summarize everything we've

1925
01:55:41,253 --> 01:55:46,983
talked about, or just the OSRA
stuff, what would be the main

1926
01:55:46,983 --> 01:55:52,427
takeaways that you hope people
come away with? start with

1927
01:55:52,427 --> 01:55:53,000
Geoff,

1928
01:55:53,000 --> 01:55:58,674
Geoff Biggs: I think, my main
takeaway for people is that the

1929
01:55:58,674 --> 01:56:04,256
OSRA looks like a big change,
but really what we're doing is

1930
01:56:04,256 --> 01:56:09,651
taking what we've done so far
and giving it, much, better.

1931
01:56:09,651 --> 01:56:14,953
Hope for the future, much
stronger, much more sustainable

1932
01:56:14,953 --> 01:56:20,721
future to ensure that we keep on
going for a very long time to

1933
01:56:20,721 --> 01:56:25,000
come. The OSRA is our route to
achieving that.

1934
01:56:25,000 --> 01:56:27,000
Audrow Nash: Hell yeah, and
Tully.

1935
01:56:27,000 --> 01:56:31,341
Tully Foote: I think my takeaway
would be that I hope that anyone

1936
01:56:31,341 --> 01:56:35,549
listening to this finds a way to
get engaged with OSRA. We have

1937
01:56:35,549 --> 01:56:39,356
opportunities for corporations.
We have opportunities for

1938
01:56:39,356 --> 01:56:43,096
individuals. I think it's
powerful for any of you to get

1939
01:56:43,096 --> 01:56:47,370
involved. We're looking to build
pathways to make that available

1940
01:56:47,370 --> 01:56:51,444
and the project will be better
if you get involved. So please

1941
01:56:51,444 --> 01:56:54,984
come out, join us, and help
build this product into a

1942
01:56:54,984 --> 01:56:58,857
greater thing. We've already got
over half a dozen members

1943
01:56:58,857 --> 01:57:02,664
committed, and I hope that if
you're at a company, please

1944
01:57:02,664 --> 01:57:04,000
consider joining us.

1945
01:57:04,000 --> 01:57:08,143
Audrow Nash: Hell yeah, and
we'll include links to all the

1946
01:57:08,143 --> 01:57:12,429
things that they might want to
look at if they want to join.

1947
01:57:12,429 --> 01:57:16,000
Awesome, so it'll all be online
for this I assume.

1948
01:57:16,000 --> 01:57:17,000
Geoff Biggs: Yes. Yes.

1949
01:57:17,000 --> 01:57:21,077
Audrow Nash: Hell yeah. Okay,
thank you both and I really, it

1950
01:57:21,077 --> 01:57:24,886
was just awesome to hear more
like the origin stories for

1951
01:57:24,886 --> 01:57:28,829
Willow Garage and Open Robotics,
like a lot of stuff that I

1952
01:57:28,829 --> 01:57:32,304
didn't know. And I think OSRA
seems like a very good

1953
01:57:32,304 --> 01:57:36,181
initiative and probably very
good for the community. And I

1954
01:57:36,181 --> 01:57:39,990
see what you're saying with
things accelerating. To robot

1955
01:57:39,990 --> 01:57:43,666
revolution, and I'm looking
forward to it. So thank you

1956
01:57:43,666 --> 01:57:44,000
both.

1957
01:57:44,000 --> 01:57:45,000
Geoff Biggs: Thank you very
much.

1958
01:57:44,000 --> 01:57:46,000
Tully Foote: Thank you.

1959
01:57:46,000 --> 01:57:50,475
Audrow Nash: Bye everyone. You
made it! What do you think? Will

1960
01:57:50,475 --> 01:57:54,880
you be joining OSRA? I'll put a
link in the description if you

1961
01:57:54,880 --> 01:57:59,071
want to check it out. One more
shout out to OSRA's founding

1962
01:57:59,071 --> 01:58:02,978
sponsors, Intrinsic, NVIDIA,
Qualcomm, Apex, Zetascale,

1963
01:58:02,978 --> 01:58:07,311
Clearpath, Ekumen, eProsima,
Picnic, Silicon Valley Robotics,

1964
01:58:07,311 --> 01:58:11,361
Canonical, and Open Navigation.
That's all for now. Happy

1965
01:58:11,361 --> 01:58:12,000
building!
