1
00:00:00,000 --> 00:00:08,469
foreign

2
00:00:00,500 --> 00:00:08,469
[Music]

3
00:00:12,799 --> 00:00:18,000
today we have Sam Bishop presenting

4
00:00:15,660 --> 00:00:20,160
opening up the final frontier expanding

5
00:00:18,000 --> 00:00:21,420
collaboration open source Aerospace

6
00:00:20,160 --> 00:00:23,460
software

7
00:00:21,420 --> 00:00:25,920
Simon's professional software developer

8
00:00:23,460 --> 00:00:27,840
amateur rocket scientist and astronomer

9
00:00:25,920 --> 00:00:29,340
after observing that Linux and open

10
00:00:27,840 --> 00:00:30,900
source are thriving in space they

11
00:00:29,340 --> 00:00:32,940
wondered why they never seen a pull

12
00:00:30,900 --> 00:00:36,120
request from SpaceX

13
00:00:32,940 --> 00:00:37,860
today Sam will discuss why this is the

14
00:00:36,120 --> 00:00:42,200
case and what we use the open source

15
00:00:37,860 --> 00:00:42,200
Community can do about it round Applause

16
00:00:45,300 --> 00:00:50,219
so yeah I'm Sam contact info as usual is

17
00:00:48,660 --> 00:00:52,379
on a slide

18
00:00:50,219 --> 00:00:55,320
um and yeah there's quite a large amount

19
00:00:52,379 --> 00:00:56,820
that goes into why this situation is

20
00:00:55,320 --> 00:01:00,180
what it is but I'll start with a bit of

21
00:00:56,820 --> 00:01:02,100
background on myself so at the start of

22
00:01:00,180 --> 00:01:04,140
covert I like a lot of people was kind

23
00:01:02,100 --> 00:01:05,640
of in a lockdown and was like what do I

24
00:01:04,140 --> 00:01:07,439
do with all this extra time when I'm not

25
00:01:05,640 --> 00:01:08,760
traveling to work and I was like okay

26
00:01:07,439 --> 00:01:10,680
maybe I'll just build like a little

27
00:01:08,760 --> 00:01:12,479
satellite I've watched these like high

28
00:01:10,680 --> 00:01:13,979
school students do it surely I can do it

29
00:01:12,479 --> 00:01:15,540
I've got way more brains than a bunch of

30
00:01:13,979 --> 00:01:16,860
high school students I've got all this

31
00:01:15,540 --> 00:01:18,960
experience they don't have and I've got

32
00:01:16,860 --> 00:01:21,659
money that they don't have I can do it

33
00:01:18,960 --> 00:01:23,400
and then I rapidly learned a lot of the

34
00:01:21,659 --> 00:01:26,340
reasons why those high school students

35
00:01:23,400 --> 00:01:28,799
can do it and I couldn't and some of

36
00:01:26,340 --> 00:01:29,880
that will be part of the talk but I

37
00:01:28,799 --> 00:01:31,619
worked out that it would actually be

38
00:01:29,880 --> 00:01:33,540
easier to start a company doing

39
00:01:31,619 --> 00:01:34,979
construction robotics in space than it

40
00:01:33,540 --> 00:01:37,140
would be to launch my own personal one

41
00:01:34,979 --> 00:01:38,880
because it turns out the whole industry

42
00:01:37,140 --> 00:01:41,040
is not designed for individuals it's

43
00:01:38,880 --> 00:01:42,180
designed for groups which is why even a

44
00:01:41,040 --> 00:01:45,000
group of high school students can do it

45
00:01:42,180 --> 00:01:46,740
but an individual will struggle so if

46
00:01:45,000 --> 00:01:48,659
any wants to talk about space robotics

47
00:01:46,740 --> 00:01:50,880
with me there's that info but I'm going

48
00:01:48,659 --> 00:01:52,380
to get on to why space is hard and it's

49
00:01:50,880 --> 00:01:54,180
not just because sometimes you drop the

50
00:01:52,380 --> 00:01:56,520
satellite and these poor Engineers

51
00:01:54,180 --> 00:01:58,979
discovered that day yes that is actually

52
00:01:56,520 --> 00:02:00,600
a expensive expensive weather satellite

53
00:01:58,979 --> 00:02:02,640
that was tipped over on its side

54
00:02:00,600 --> 00:02:04,500
accidentally and they actually fixed it

55
00:02:02,640 --> 00:02:06,360
amazingly enough it was cheaper to fix

56
00:02:04,500 --> 00:02:09,479
than rebuild but

57
00:02:06,360 --> 00:02:11,520
so the actual hard Parts not the fun

58
00:02:09,479 --> 00:02:12,720
parts of dropping things is that you

59
00:02:11,520 --> 00:02:14,099
actually have to deal with a lot of

60
00:02:12,720 --> 00:02:15,660
physical challenges in the space

61
00:02:14,099 --> 00:02:17,940
environment and they have some logical

62
00:02:15,660 --> 00:02:19,379
effects and some Hardware effects and

63
00:02:17,940 --> 00:02:21,360
since I'm covering open source in the

64
00:02:19,379 --> 00:02:22,860
general sense both hardware and software

65
00:02:21,360 --> 00:02:24,239
I wanted to cover some of the things

66
00:02:22,860 --> 00:02:27,000
that are relevant in the hardware

67
00:02:24,239 --> 00:02:28,379
context as well so things like how in

68
00:02:27,000 --> 00:02:30,900
space you have no thermal management

69
00:02:28,379 --> 00:02:32,040
other than what you can radiate away so

70
00:02:30,900 --> 00:02:33,660
you actually have to be very careful

71
00:02:32,040 --> 00:02:34,920
about how much heat you produce which

72
00:02:33,660 --> 00:02:37,200
means you've got to pay attention to all

73
00:02:34,920 --> 00:02:38,819
of your thermals very very religiously

74
00:02:37,200 --> 00:02:40,500
you've got to be careful that you don't

75
00:02:38,819 --> 00:02:42,480
make your circuit boards out of Coatings

76
00:02:40,500 --> 00:02:44,580
that contain gases that could outgass

77
00:02:42,480 --> 00:02:46,560
and destroy your instruments or gas

78
00:02:44,580 --> 00:02:49,200
bubbles that could cause traces on the

79
00:02:46,560 --> 00:02:50,340
pcbs to lift up and short-circuit and

80
00:02:49,200 --> 00:02:52,500
all of these kinds of things come

81
00:02:50,340 --> 00:02:55,860
together in a way that makes these extra

82
00:02:52,500 --> 00:02:57,840
details important and then you get more

83
00:02:55,860 --> 00:02:59,940
obvious stuff like radiation well yeah

84
00:02:57,840 --> 00:03:02,160
radiation is pretty easy to deal with

85
00:02:59,940 --> 00:03:04,920
you just have to have redundancy and

86
00:03:02,160 --> 00:03:06,959
redundancy and redundancy So Physical

87
00:03:04,920 --> 00:03:08,819
redundancy consensus algorithms giving

88
00:03:06,959 --> 00:03:11,280
you logical redundancy and then

89
00:03:08,819 --> 00:03:13,680
reprogrammable Hardware like fpgas and

90
00:03:11,280 --> 00:03:15,659
cplds to give you extra Hardware

91
00:03:13,680 --> 00:03:18,959
redundancy but controlled by software

92
00:03:15,659 --> 00:03:21,540
because if a heavy ion strikes into your

93
00:03:18,959 --> 00:03:23,879
circuit and completely destroys a gate

94
00:03:21,540 --> 00:03:25,680
well do you want that to be completely

95
00:03:23,879 --> 00:03:27,120
hosed no you'd like to be able to do

96
00:03:25,680 --> 00:03:28,860
something with the rest of the millions

97
00:03:27,120 --> 00:03:31,019
of transistors in that expensive

98
00:03:28,860 --> 00:03:32,940
component so things like having the

99
00:03:31,019 --> 00:03:34,500
ability to reprogram the rest of that

100
00:03:32,940 --> 00:03:36,420
silicon device that spent hundreds of

101
00:03:34,500 --> 00:03:37,860
thousands of dollars buying is kind of

102
00:03:36,420 --> 00:03:40,319
valuable which is why we see a lot of

103
00:03:37,860 --> 00:03:41,879
fpgas in space and then there's the

104
00:03:40,319 --> 00:03:44,220
communication challenges of being in

105
00:03:41,879 --> 00:03:46,379
space you can't just keep an SSH

106
00:03:44,220 --> 00:03:48,599
terminal open to your satellite well you

107
00:03:46,379 --> 00:03:50,879
can but you need a very big budget for

108
00:03:48,599 --> 00:03:52,140
that and if you don't have that massive

109
00:03:50,879 --> 00:03:54,120
budget you'll be dealing with things

110
00:03:52,140 --> 00:03:56,099
like communication delays and

111
00:03:54,120 --> 00:03:57,900
communication windows and orbital

112
00:03:56,099 --> 00:03:59,340
phasing meaning that sometimes you may

113
00:03:57,900 --> 00:04:01,440
only have on a very cheap Mission

114
00:03:59,340 --> 00:04:03,659
something like 30 minutes to an hour

115
00:04:01,440 --> 00:04:06,180
total to communicate with your satellite

116
00:04:03,659 --> 00:04:09,599
anywhere between once a day and several

117
00:04:06,180 --> 00:04:11,519
weeks so you might have to work out your

118
00:04:09,599 --> 00:04:14,280
software so that you can send a batch of

119
00:04:11,519 --> 00:04:17,519
updates nicely compressed in order to

120
00:04:14,280 --> 00:04:19,139
all six weeks of the next set of charms

121
00:04:17,519 --> 00:04:20,940
out to that satellite

122
00:04:19,139 --> 00:04:23,520
and then wait for the results in six

123
00:04:20,940 --> 00:04:25,139
weeks time and then you might decide

124
00:04:23,520 --> 00:04:26,759
okay well I can do that thing up but I

125
00:04:25,139 --> 00:04:27,720
need more data down so I'll muck around

126
00:04:26,759 --> 00:04:29,340
and I'll work with something like

127
00:04:27,720 --> 00:04:30,900
Optical or laser Communications on the

128
00:04:29,340 --> 00:04:32,820
way down because it's easy to go down

129
00:04:30,900 --> 00:04:34,259
than it is to go back up because you

130
00:04:32,820 --> 00:04:36,540
have a hard time putting very large

131
00:04:34,259 --> 00:04:38,520
receiving telescopes on the bottom of

132
00:04:36,540 --> 00:04:39,780
small cheap satellites but they could

133
00:04:38,520 --> 00:04:42,900
put lasers really easily on there

134
00:04:39,780 --> 00:04:44,040
because lasers can be nice and small but

135
00:04:42,900 --> 00:04:45,120
other things that are coming down the

136
00:04:44,040 --> 00:04:46,259
pipe are satellite to satellite

137
00:04:45,120 --> 00:04:47,820
communication people have done

138
00:04:46,259 --> 00:04:49,620
experiments with using like iridium

139
00:04:47,820 --> 00:04:51,660
modems so like you'd get with a SAT

140
00:04:49,620 --> 00:04:53,160
phone you can just buy they put that on

141
00:04:51,660 --> 00:04:55,500
their little satellite and then they

142
00:04:53,160 --> 00:04:56,940
actually have a very low data rate but a

143
00:04:55,500 --> 00:04:58,020
consistent one and that gives them

144
00:04:56,940 --> 00:05:00,240
another way to get around this problem

145
00:04:58,020 --> 00:05:01,919
it's great for these persistent

146
00:05:00,240 --> 00:05:04,380
connectivity scenarios but this Hardware

147
00:05:01,919 --> 00:05:06,419
isn't widely available because despite

148
00:05:04,380 --> 00:05:08,280
being fairly standard Hardware it's not

149
00:05:06,419 --> 00:05:09,960
open it's the behest of a company with

150
00:05:08,280 --> 00:05:12,360
the licensing and a lot of radio

151
00:05:09,960 --> 00:05:14,160
regulations which is one of the things

152
00:05:12,360 --> 00:05:15,720
that I learned in that process of why I

153
00:05:14,160 --> 00:05:17,100
can't do it individually radio

154
00:05:15,720 --> 00:05:18,979
regulations are not designed for

155
00:05:17,100 --> 00:05:21,479
individuals when it comes to Aerospace

156
00:05:18,979 --> 00:05:23,220
then there's the power which comes back

157
00:05:21,479 --> 00:05:26,100
to The Thermals but the power is the

158
00:05:23,220 --> 00:05:27,900
driver and it's one heavy solar panels

159
00:05:26,100 --> 00:05:30,600
are weight and batteries are even

160
00:05:27,900 --> 00:05:33,000
heavier and you'll find that even on a

161
00:05:30,600 --> 00:05:35,820
professional large satellite you will be

162
00:05:33,000 --> 00:05:38,400
dealing with a significant percentage of

163
00:05:35,820 --> 00:05:40,560
the mass that isn't fuel will be your

164
00:05:38,400 --> 00:05:42,120
power because if you want to do anything

165
00:05:40,560 --> 00:05:44,580
useful you've got to have the

166
00:05:42,120 --> 00:05:46,680
electricity to do it and then when

167
00:05:44,580 --> 00:05:48,240
you're not necessarily going to have Sun

168
00:05:46,680 --> 00:05:49,979
such as if you have to turn the

169
00:05:48,240 --> 00:05:52,380
satellite to point The Big Dish at

170
00:05:49,979 --> 00:05:53,940
something well then you need batteries

171
00:05:52,380 --> 00:05:56,039
to have enough power to do what you're

172
00:05:53,940 --> 00:05:57,360
doing and particularly with radio that

173
00:05:56,039 --> 00:05:58,860
becomes important if you need a high

174
00:05:57,360 --> 00:06:01,560
bandwidth link to send down big pretty

175
00:05:58,860 --> 00:06:03,900
pictures you might have to aim that dish

176
00:06:01,560 --> 00:06:05,880
or that antenna and if you're aiming it

177
00:06:03,900 --> 00:06:07,560
that means that your

178
00:06:05,880 --> 00:06:08,880
solar panels will not necessarily be in

179
00:06:07,560 --> 00:06:10,680
the right place depending on the time of

180
00:06:08,880 --> 00:06:12,479
day with all of those previously

181
00:06:10,680 --> 00:06:14,460
mentioned orbital phasing timing things

182
00:06:12,479 --> 00:06:16,979
which means you might need a really

183
00:06:14,460 --> 00:06:18,240
decent burst of battery power why once

184
00:06:16,979 --> 00:06:20,820
every six months but if you don't have

185
00:06:18,240 --> 00:06:22,919
it your satellite dies so you have to

186
00:06:20,820 --> 00:06:25,620
pay attention to all of these things and

187
00:06:22,919 --> 00:06:27,120
as nicely available as the batteries are

188
00:06:25,620 --> 00:06:30,300
getting you can just literally use like

189
00:06:27,120 --> 00:06:31,979
the the 18650s for a lot of the smaller

190
00:06:30,300 --> 00:06:33,979
satellites and they do get widely used

191
00:06:31,979 --> 00:06:36,600
for cubesats and sort of project SATs

192
00:06:33,979 --> 00:06:38,000
that lithium-ion technology is now quite

193
00:06:36,600 --> 00:06:40,380
well accepted

194
00:06:38,000 --> 00:06:41,400
as well as the fact that Electronics

195
00:06:40,380 --> 00:06:43,259
getting better and better all the time

196
00:06:41,400 --> 00:06:45,240
means that we can use thermal management

197
00:06:43,259 --> 00:06:46,500
and power management lower States and

198
00:06:45,240 --> 00:06:48,780
all of the great stuff that we get for

199
00:06:46,500 --> 00:06:50,160
these tiny embedded controllers to save

200
00:06:48,780 --> 00:06:51,780
every watch that we don't want to use

201
00:06:50,160 --> 00:06:54,660
when we run our tiny projects on coin

202
00:06:51,780 --> 00:06:55,740
cells which is great but at the same

203
00:06:54,660 --> 00:06:58,080
time

204
00:06:55,740 --> 00:07:00,180
with a project this much stuff needing

205
00:06:58,080 --> 00:07:01,740
to be checked off and cared for you

206
00:07:00,180 --> 00:07:04,020
don't want to repeat yourself

207
00:07:01,740 --> 00:07:06,780
that's a diagram of the logical system

208
00:07:04,020 --> 00:07:09,000
part of The Logical system from one of

209
00:07:06,780 --> 00:07:11,220
NASA's open source projects that looks

210
00:07:09,000 --> 00:07:13,680
after a satellite software not even the

211
00:07:11,220 --> 00:07:15,000
hardware just the software and all of

212
00:07:13,680 --> 00:07:17,580
those pieces interact with different

213
00:07:15,000 --> 00:07:19,620
bits of hardware subsystems and

214
00:07:17,580 --> 00:07:22,500
as a developer I look at a diagram like

215
00:07:19,620 --> 00:07:24,900
that and I never want to do that twice

216
00:07:22,500 --> 00:07:26,759
if I build that for myself

217
00:07:24,900 --> 00:07:28,680
I never want to do that again because

218
00:07:26,759 --> 00:07:30,599
that is a significant amount of time and

219
00:07:28,680 --> 00:07:32,819
investment that is operating system

220
00:07:30,599 --> 00:07:35,400
levels of time and investment and while

221
00:07:32,819 --> 00:07:36,599
you know a Linux conference an open

222
00:07:35,400 --> 00:07:38,220
source conference we have lots of

223
00:07:36,599 --> 00:07:40,860
example operating systems to point to

224
00:07:38,220 --> 00:07:42,419
that individuals have done there are not

225
00:07:40,860 --> 00:07:44,520
many individuals who have done that

226
00:07:42,419 --> 00:07:46,620
multiple times

227
00:07:44,520 --> 00:07:49,139
because it does represent that level of

228
00:07:46,620 --> 00:07:51,300
investment that a person does not want

229
00:07:49,139 --> 00:07:52,979
to do alone repeatedly it is bigger than

230
00:07:51,300 --> 00:07:54,539
one person to do this kind of software

231
00:07:52,979 --> 00:07:56,819
correctly and to do the hardware

232
00:07:54,539 --> 00:07:59,099
correctly absolutely shouldn't be done

233
00:07:56,819 --> 00:08:00,539
alone and that gets into the height of

234
00:07:59,099 --> 00:08:02,099
challenges about this being a

235
00:08:00,539 --> 00:08:04,380
environment where you have to work with

236
00:08:02,099 --> 00:08:05,400
other people and those are the harder

237
00:08:04,380 --> 00:08:07,680
challenges in this kind of an

238
00:08:05,400 --> 00:08:09,479
environment for open source it's the the

239
00:08:07,680 --> 00:08:11,460
Social Challenges it's the things like

240
00:08:09,479 --> 00:08:13,440
how in a traditional context an

241
00:08:11,460 --> 00:08:14,940
aerospace company has a lot of layers of

242
00:08:13,440 --> 00:08:16,380
management there will be managers and

243
00:08:14,940 --> 00:08:18,060
managers and Engineering managers and

244
00:08:16,380 --> 00:08:20,460
Engineering review managers and process

245
00:08:18,060 --> 00:08:22,860
control managers and managers to manage

246
00:08:20,460 --> 00:08:24,840
the process control the supply chain and

247
00:08:22,860 --> 00:08:26,580
and all of the layers up and down and it

248
00:08:24,840 --> 00:08:30,000
makes sense once you know how many

249
00:08:26,580 --> 00:08:31,560
things go into a large project but these

250
00:08:30,000 --> 00:08:33,839
companies truly do live and die by this

251
00:08:31,560 --> 00:08:36,659
kind of process control because

252
00:08:33,839 --> 00:08:38,880
they're building gigantic things and if

253
00:08:36,659 --> 00:08:40,500
someone who's outside the organization

254
00:08:38,880 --> 00:08:42,659
say they open source their flight

255
00:08:40,500 --> 00:08:44,940
software and anyone in the world could

256
00:08:42,659 --> 00:08:46,380
contribute to that flight software well

257
00:08:44,940 --> 00:08:47,700
how are they going to stop someone in

258
00:08:46,380 --> 00:08:49,560
the wrong Department who knows nothing

259
00:08:47,700 --> 00:08:51,839
about it messing up what they're doing

260
00:08:49,560 --> 00:08:53,339
for their project let alone someone on

261
00:08:51,839 --> 00:08:54,600
the other side of the world they don't

262
00:08:53,339 --> 00:08:55,860
want their drone falling out of the air

263
00:08:54,600 --> 00:08:58,440
they don't want their plane falling out

264
00:08:55,860 --> 00:09:00,300
of the air they want to make sure that

265
00:08:58,440 --> 00:09:01,500
that software is fully controlled and it

266
00:09:00,300 --> 00:09:03,180
gets harder to deal with the people

267
00:09:01,500 --> 00:09:04,920
inside the organization because you

268
00:09:03,180 --> 00:09:07,320
can't just say oh every commit from

269
00:09:04,920 --> 00:09:09,660
someone who doesn't have a whatever.com

270
00:09:07,320 --> 00:09:13,019
email address has to be double checked

271
00:09:09,660 --> 00:09:15,420
well the rest of your stuff might have

272
00:09:13,019 --> 00:09:17,580
the same domain name the rest of the

273
00:09:15,420 --> 00:09:19,140
staff might not know what they're doing

274
00:09:17,580 --> 00:09:21,480
but think they know what they're doing

275
00:09:19,140 --> 00:09:23,279
and put commits in there is management

276
00:09:21,480 --> 00:09:25,320
complexity that these companies are

277
00:09:23,279 --> 00:09:27,480
afraid of and are not necessarily

278
00:09:25,320 --> 00:09:29,580
prepared to take on to manage open

279
00:09:27,480 --> 00:09:32,279
source in their kind of an environment

280
00:09:29,580 --> 00:09:34,140
and it really does come down to the

281
00:09:32,279 --> 00:09:37,019
numbers because if you look at something

282
00:09:34,140 --> 00:09:38,100
like a 747-400 so that's from 1988 was

283
00:09:37,019 --> 00:09:39,959
the first time that got off the ground

284
00:09:38,100 --> 00:09:42,420
and that had four hundred thousand lines

285
00:09:39,959 --> 00:09:45,540
of code in its flight Control software

286
00:09:42,420 --> 00:09:48,600
city is in the future we're on the 777

287
00:09:45,540 --> 00:09:51,660
it takes off being flown by two and a

288
00:09:48,600 --> 00:09:54,720
half million lines of software

289
00:09:51,660 --> 00:09:56,700
and that's mid 90s it has definitely

290
00:09:54,720 --> 00:09:59,640
gotten bigger it has definitely gotten

291
00:09:56,700 --> 00:10:01,080
more complex there are little niches and

292
00:09:59,640 --> 00:10:02,820
places where these systems interact

293
00:10:01,080 --> 00:10:05,399
between layers and air gaps and things

294
00:10:02,820 --> 00:10:08,399
like well how do I put the where is the

295
00:10:05,399 --> 00:10:10,740
plane now info into a system so that the

296
00:10:08,399 --> 00:10:13,380
really lower security environment of the

297
00:10:10,740 --> 00:10:14,820
iPad in the back of the plane so that

298
00:10:13,380 --> 00:10:17,160
someone can watch their you know movie

299
00:10:14,820 --> 00:10:19,680
for three hours can still see where the

300
00:10:17,160 --> 00:10:21,360
plane is on a pretty map still how do

301
00:10:19,680 --> 00:10:24,000
they render that out of the navigation

302
00:10:21,360 --> 00:10:26,000
software that you don't want someone

303
00:10:24,000 --> 00:10:28,440
being able to hack from the plane Wi-Fi

304
00:10:26,000 --> 00:10:30,600
there are complexities in these that

305
00:10:28,440 --> 00:10:32,459
warrant this level of management

306
00:10:30,600 --> 00:10:35,640
but it can be improved for an open

307
00:10:32,459 --> 00:10:37,620
context and that is part of the economic

308
00:10:35,640 --> 00:10:39,480
problem there is an economic cost in

309
00:10:37,620 --> 00:10:42,180
moving forward to that kind of a change

310
00:10:39,480 --> 00:10:44,880
you need to accept that those costs will

311
00:10:42,180 --> 00:10:46,980
both have benefits and negatives there

312
00:10:44,880 --> 00:10:49,500
will be risks things like if you're a

313
00:10:46,980 --> 00:10:52,200
defense contractor selling support for a

314
00:10:49,500 --> 00:10:54,180
proprietary package to a military well

315
00:10:52,200 --> 00:10:55,079
if that becomes open software you aren't

316
00:10:54,180 --> 00:10:56,279
going to be able to sell it anymore

317
00:10:55,079 --> 00:10:58,500
they're just going to be able to get it

318
00:10:56,279 --> 00:11:00,720
so you can't have that kind of a thing

319
00:10:58,500 --> 00:11:03,060
anymore in an open source

320
00:11:00,720 --> 00:11:04,860
environment that way you have to write

321
00:11:03,060 --> 00:11:07,260
that Revenue off and focus on other

322
00:11:04,860 --> 00:11:10,019
things or you cannot open source that

323
00:11:07,260 --> 00:11:11,880
software and things stay the same and

324
00:11:10,019 --> 00:11:15,000
there's a risk of the competitors that

325
00:11:11,880 --> 00:11:16,800
they work with so if you have you know

326
00:11:15,000 --> 00:11:18,420
Boeing Lockheed and all of these

327
00:11:16,800 --> 00:11:21,120
companies they compete quite viciously

328
00:11:18,420 --> 00:11:22,680
for contracts and they are genuinely

329
00:11:21,120 --> 00:11:24,779
going to be concerned that if they

330
00:11:22,680 --> 00:11:26,700
publish their own work openly their

331
00:11:24,779 --> 00:11:28,620
competitors will take that work undercut

332
00:11:26,700 --> 00:11:30,420
their bid and put them out of business

333
00:11:28,620 --> 00:11:31,860
they don't want to have that and that

334
00:11:30,420 --> 00:11:33,720
gets more and more aggressive the

335
00:11:31,860 --> 00:11:34,980
smaller these companies get when you get

336
00:11:33,720 --> 00:11:36,959
down to the kind of small Innovative

337
00:11:34,980 --> 00:11:39,360
companies working on like small Man

338
00:11:36,959 --> 00:11:42,480
portable drone systems for you know

339
00:11:39,360 --> 00:11:44,399
inspection or you know cattle checking

340
00:11:42,480 --> 00:11:46,560
these these ones are very very

341
00:11:44,399 --> 00:11:48,540
competitive and probably leveraging open

342
00:11:46,560 --> 00:11:50,220
source more than the bigger ones do but

343
00:11:48,540 --> 00:11:51,839
at the same time the fear is there and

344
00:11:50,220 --> 00:11:54,060
they will be leveraging it one way it

345
00:11:51,839 --> 00:11:56,880
goes in it does not come back out

346
00:11:54,060 --> 00:11:58,920
and part of that is regulatory

347
00:11:56,880 --> 00:12:01,399
this is more of an issue with space than

348
00:11:58,920 --> 00:12:04,019
air than the Aero part specifically

349
00:12:01,399 --> 00:12:06,120
every country generally speaking has

350
00:12:04,019 --> 00:12:07,740
some kind of export controls for the US

351
00:12:06,120 --> 00:12:09,899
it's itas or the international traffic

352
00:12:07,740 --> 00:12:11,940
and arms regulations the UK has the

353
00:12:09,899 --> 00:12:13,440
export Control Act the Australian

354
00:12:11,940 --> 00:12:16,380
government has the defense trade

355
00:12:13,440 --> 00:12:18,300
controls act and while some of this does

356
00:12:16,380 --> 00:12:20,040
not apply to open source because in

357
00:12:18,300 --> 00:12:22,740
particular America have done a good job

358
00:12:20,040 --> 00:12:24,120
of clearly cutting open source out you

359
00:12:22,740 --> 00:12:25,800
know there is a relatively

360
00:12:24,120 --> 00:12:28,079
straightforward way to determine if it

361
00:12:25,800 --> 00:12:31,740
is or is not under these for the case of

362
00:12:28,079 --> 00:12:33,060
software there is ambiguity here well

363
00:12:31,740 --> 00:12:34,860
covered in the paper that I actually

364
00:12:33,060 --> 00:12:37,320
pulled that slide from referenced in the

365
00:12:34,860 --> 00:12:39,600
top corner there and that regulatory

366
00:12:37,320 --> 00:12:40,980
ambiguity is where things get dicey for

367
00:12:39,600 --> 00:12:43,500
a company that wants to try and publish

368
00:12:40,980 --> 00:12:44,820
their open source software uh the state

369
00:12:43,500 --> 00:12:46,380
department have a lovely definition

370
00:12:44,820 --> 00:12:48,240
because they are the ones who have their

371
00:12:46,380 --> 00:12:51,000
directorate of trade control that looks

372
00:12:48,240 --> 00:12:53,160
after this and that is that itar has a

373
00:12:51,000 --> 00:12:55,860
list of services and articles and

374
00:12:53,160 --> 00:12:58,440
related technical data that are the

375
00:12:55,860 --> 00:13:00,600
defense articles and then that technical

376
00:12:58,440 --> 00:13:02,820
data definition includes software

377
00:13:00,600 --> 00:13:04,079
directly related to defense articles

378
00:13:02,820 --> 00:13:06,660
creating a nice little circular

379
00:13:04,079 --> 00:13:09,420
reference which means that when time

380
00:13:06,660 --> 00:13:12,360
comes and you have this stuff assessed

381
00:13:09,420 --> 00:13:14,639
to is this or isn't this suitable for

382
00:13:12,360 --> 00:13:16,560
inclusion well

383
00:13:14,639 --> 00:13:18,060
you actually have to have a human go

384
00:13:16,560 --> 00:13:20,040
through it you can't just say for

385
00:13:18,060 --> 00:13:21,660
certain because there is this circular

386
00:13:20,040 --> 00:13:23,399
Loop of world does it come under it

387
00:13:21,660 --> 00:13:25,860
under this interpretation back through

388
00:13:23,399 --> 00:13:27,899
the chain or well when we've applied it

389
00:13:25,860 --> 00:13:29,880
to the thing that it's not used on it's

390
00:13:27,899 --> 00:13:32,880
technically a input for something that

391
00:13:29,880 --> 00:13:35,220
is and that makes it more or less and

392
00:13:32,880 --> 00:13:39,360
it's very ambiguous very manual very

393
00:13:35,220 --> 00:13:40,560
human interpreted over and over and that

394
00:13:39,360 --> 00:13:43,740
is where a big part of the risk for

395
00:13:40,560 --> 00:13:46,139
these companies is and one particularly

396
00:13:43,740 --> 00:13:47,639
annoying little detail is that itire has

397
00:13:46,139 --> 00:13:50,160
a process for marking something as

398
00:13:47,639 --> 00:13:52,500
public so that that half of there where

399
00:13:50,160 --> 00:13:54,720
it's exempt and you don't have to worry

400
00:13:52,500 --> 00:13:58,040
about itar restrictions on your lovely

401
00:13:54,720 --> 00:14:00,420
project but crucially of the eight

402
00:13:58,040 --> 00:14:02,399
mandated ways to make it publicly

403
00:14:00,420 --> 00:14:04,500
available and publicly available is a

404
00:14:02,399 --> 00:14:06,120
noun in this context

405
00:14:04,500 --> 00:14:07,019
putting it on the internet is not one of

406
00:14:06,120 --> 00:14:09,180
them

407
00:14:07,019 --> 00:14:11,160
so you can't just publish the code to

408
00:14:09,180 --> 00:14:13,920
GitHub and have that pull request count

409
00:14:11,160 --> 00:14:16,139
as publishing to the public access in a

410
00:14:13,920 --> 00:14:17,160
way that means it's itar free no you

411
00:14:16,139 --> 00:14:19,079
actually have to have a gatekeeping

412
00:14:17,160 --> 00:14:20,639
mechanism some kind of an extra step in

413
00:14:19,079 --> 00:14:23,519
there do one of those other eight things

414
00:14:20,639 --> 00:14:25,139
first and then it can be public

415
00:14:23,519 --> 00:14:27,540
otherwise you run the risk of breaching

416
00:14:25,139 --> 00:14:28,980
itar which is the kind of thing that you

417
00:14:27,540 --> 00:14:30,899
do not want to do if you want your

418
00:14:28,980 --> 00:14:33,120
company to stay alive if you aren't say

419
00:14:30,899 --> 00:14:34,680
Boeing because they can get away with it

420
00:14:33,120 --> 00:14:37,380
with a bit of a slap on the wrist and a

421
00:14:34,680 --> 00:14:39,660
large large fine but if you're you know

422
00:14:37,380 --> 00:14:41,579
trying to make drones and you breach

423
00:14:39,660 --> 00:14:43,500
that because someone nicked your

424
00:14:41,579 --> 00:14:45,779
software and you didn't do all the right

425
00:14:43,500 --> 00:14:47,760
dotting of eyes and crossing of T's well

426
00:14:45,779 --> 00:14:49,440
you you might not just walk away with no

427
00:14:47,760 --> 00:14:52,560
company you might walk away with jail

428
00:14:49,440 --> 00:14:54,300
time they take that pretty seriously

429
00:14:52,560 --> 00:14:56,160
um and part of that is where we get to

430
00:14:54,300 --> 00:14:59,639
the last part of this the liability in

431
00:14:56,160 --> 00:15:01,380
general risk the liability is a huge

432
00:14:59,639 --> 00:15:04,740
factor for these companies these

433
00:15:01,380 --> 00:15:07,740
companies are very very embedded in a

434
00:15:04,740 --> 00:15:10,260
liability World they are that is you

435
00:15:07,740 --> 00:15:12,779
know they build planes they people's

436
00:15:10,260 --> 00:15:15,540
lives are in their hands and and that

437
00:15:12,779 --> 00:15:17,160
matters to both them and to the people

438
00:15:15,540 --> 00:15:19,380
who make the laws that they have to work

439
00:15:17,160 --> 00:15:22,440
within and those kinds of liability

440
00:15:19,380 --> 00:15:25,079
risks are legal risks to the companies

441
00:15:22,440 --> 00:15:27,420
lawyers they they are unproven legal

442
00:15:25,079 --> 00:15:29,100
risks they are untested by precedent

443
00:15:27,420 --> 00:15:31,680
there is no case law for them to point

444
00:15:29,100 --> 00:15:34,079
to when they are asked a question is it

445
00:15:31,680 --> 00:15:35,880
safe for us to whatever the lawyer's

446
00:15:34,079 --> 00:15:38,339
answer will be I have nothing to say it

447
00:15:35,880 --> 00:15:40,320
isn't therefore it is not safe

448
00:15:38,339 --> 00:15:43,019
and there will be a very big long

449
00:15:40,320 --> 00:15:45,360
process to try and change that

450
00:15:43,019 --> 00:15:46,920
can be improved and it has been improved

451
00:15:45,360 --> 00:15:50,399
in some places in other Industries

452
00:15:46,920 --> 00:15:53,279
already and that legal liability is part

453
00:15:50,399 --> 00:15:55,380
of the reason that open source licenses

454
00:15:53,279 --> 00:15:58,199
in general are being used in a one-way

455
00:15:55,380 --> 00:16:00,360
fashion the code goes in they review it

456
00:15:58,199 --> 00:16:01,079
to their own internal satisfaction they

457
00:16:00,360 --> 00:16:02,699
can

458
00:16:01,079 --> 00:16:04,500
basically look at that little bit at the

459
00:16:02,699 --> 00:16:07,079
bottom of the MIT license that says no

460
00:16:04,500 --> 00:16:09,959
warranty no liability if you screw up

461
00:16:07,079 --> 00:16:12,300
it's on you that part they take the

462
00:16:09,959 --> 00:16:14,279
burden off and all the work they do

463
00:16:12,300 --> 00:16:16,380
internally to take the burden of that

464
00:16:14,279 --> 00:16:18,899
is valuable to the company

465
00:16:16,380 --> 00:16:21,300
it's them doing hard work to accept the

466
00:16:18,899 --> 00:16:22,920
liability and that's part of their value

467
00:16:21,300 --> 00:16:26,100
proposition in using open source

468
00:16:22,920 --> 00:16:28,320
software and to open things up more they

469
00:16:26,100 --> 00:16:31,560
have to contribute back that work in

470
00:16:28,320 --> 00:16:34,260
part even if not in full but in part by

471
00:16:31,560 --> 00:16:36,720
way of trust that they actually use that

472
00:16:34,260 --> 00:16:39,180
code if they publish what they use it

473
00:16:36,720 --> 00:16:41,579
means that that code was actually

474
00:16:39,180 --> 00:16:44,279
verified good enough

475
00:16:41,579 --> 00:16:47,459
and that creates a moral implication if

476
00:16:44,279 --> 00:16:48,899
not a legal implication and that's where

477
00:16:47,459 --> 00:16:50,220
things get more complex for the

478
00:16:48,899 --> 00:16:51,899
companies and it becomes more of a

479
00:16:50,220 --> 00:16:53,639
quagmire for them to try and get dragged

480
00:16:51,899 --> 00:16:56,279
out of by the open Community because

481
00:16:53,639 --> 00:16:57,959
it's very difficult to avoid the risks

482
00:16:56,279 --> 00:17:00,180
of them winding up in a crab mentality

483
00:16:57,959 --> 00:17:01,800
if you make a small light aircraft and

484
00:17:00,180 --> 00:17:04,140
you borrow the code for the navigation

485
00:17:01,800 --> 00:17:04,919
display and a pilot flies it into a

486
00:17:04,140 --> 00:17:07,380
mountain

487
00:17:04,919 --> 00:17:09,240
well the people who put the code in

488
00:17:07,380 --> 00:17:11,100
there they're not actually responsible

489
00:17:09,240 --> 00:17:12,480
for you making sure that that won't fly

490
00:17:11,100 --> 00:17:14,760
someone into a mountain because the GPS

491
00:17:12,480 --> 00:17:17,640
is backwards you are because you made

492
00:17:14,760 --> 00:17:19,620
the plane you put the code in but

493
00:17:17,640 --> 00:17:21,179
you can very easily if you want to avoid

494
00:17:19,620 --> 00:17:22,740
your own legal liability your own

495
00:17:21,179 --> 00:17:24,240
lawyers probably recommending this to

496
00:17:22,740 --> 00:17:26,880
you every time you ask them something

497
00:17:24,240 --> 00:17:28,679
they will be like blame everyone you can

498
00:17:26,880 --> 00:17:30,840
other than yourself because that spreads

499
00:17:28,679 --> 00:17:32,880
the blame and that helps your case and

500
00:17:30,840 --> 00:17:35,340
so they will probably say it may have

501
00:17:32,880 --> 00:17:37,559
been in people who gave us code

502
00:17:35,340 --> 00:17:39,360
and you get the crab mentality you get

503
00:17:37,559 --> 00:17:41,580
one company dragging the other into the

504
00:17:39,360 --> 00:17:44,520
bucket again and everyone looks at the

505
00:17:41,580 --> 00:17:46,860
code and goes we can't trust this it's

506
00:17:44,520 --> 00:17:48,840
not good enough we shouldn't do it this

507
00:17:46,860 --> 00:17:51,660
way because look at that thing that just

508
00:17:48,840 --> 00:17:54,059
happened it taints the entire perception

509
00:17:51,660 --> 00:17:56,539
around using open code in this kind of

510
00:17:54,059 --> 00:18:00,120
an environment where safety is critical

511
00:17:56,539 --> 00:18:02,640
liability is absolutely absolutely lives

512
00:18:00,120 --> 00:18:04,500
are on the line level of liability and

513
00:18:02,640 --> 00:18:07,200
it is going to take time for us to

514
00:18:04,500 --> 00:18:09,480
overcome this but

515
00:18:07,200 --> 00:18:11,340
in the long run it's better because

516
00:18:09,480 --> 00:18:13,200
I kind of don't want to die in a plane

517
00:18:11,340 --> 00:18:14,700
crash because some really smart person

518
00:18:13,200 --> 00:18:17,520
worked out how to mess with the flight

519
00:18:14,700 --> 00:18:19,020
controls by jumping at air gap on the

520
00:18:17,520 --> 00:18:20,400
Wi-Fi from the plane to the navigation

521
00:18:19,020 --> 00:18:21,980
systems because he worked out he could

522
00:18:20,400 --> 00:18:25,200
use some length of wire with as a radio

523
00:18:21,980 --> 00:18:27,900
like I see air gap breaches come through

524
00:18:25,200 --> 00:18:29,520
on the little fun list of random hacks

525
00:18:27,900 --> 00:18:31,200
that I get shown by some people I you

526
00:18:29,520 --> 00:18:32,820
know follow and I'm like wow that's

527
00:18:31,200 --> 00:18:36,000
insanely clever I can't believe you did

528
00:18:32,820 --> 00:18:39,419
that with the hard drive speed noise or

529
00:18:36,000 --> 00:18:41,400
the PC speaker or the LED light and a

530
00:18:39,419 --> 00:18:42,960
camera somewhere else in the room the

531
00:18:41,400 --> 00:18:46,799
amount of ways you can jump into your

532
00:18:42,960 --> 00:18:48,419
Gap just defies belief and when your

533
00:18:46,799 --> 00:18:50,400
entire security model consists of well

534
00:18:48,419 --> 00:18:52,679
we didn't link these two directly so

535
00:18:50,400 --> 00:18:55,620
surely that's secure enough

536
00:18:52,679 --> 00:18:56,760
I kind of don't like that at the back of

537
00:18:55,620 --> 00:18:58,140
my mind in a part that I just try and

538
00:18:56,760 --> 00:19:00,419
switch off every time I bought a plane

539
00:18:58,140 --> 00:19:01,919
and it would be nice to know that more

540
00:19:00,419 --> 00:19:04,740
people were looking at this that that

541
00:19:01,919 --> 00:19:06,720
code was being opened up that we are

542
00:19:04,740 --> 00:19:08,520
actually making those bugs shallow that

543
00:19:06,720 --> 00:19:10,740
we are answering the classic why of what

544
00:19:08,520 --> 00:19:13,140
makes open source software worth it it's

545
00:19:10,740 --> 00:19:14,820
that many eyes make all bugs shallow it

546
00:19:13,140 --> 00:19:18,840
makes the software better because it's

547
00:19:14,820 --> 00:19:20,340
not just the one person doing it and we

548
00:19:18,840 --> 00:19:22,500
have an example of where this has been

549
00:19:20,340 --> 00:19:25,020
done before we've got Automotive grade

550
00:19:22,500 --> 00:19:26,760
Linux now Automotive grade Linux was

551
00:19:25,020 --> 00:19:29,100
started in 2012 so it's 10-year

552
00:19:26,760 --> 00:19:30,780
anniversary last year and it's a

553
00:19:29,100 --> 00:19:32,039
collaborative project that drags in a

554
00:19:30,780 --> 00:19:34,260
bunch of the automotive companies from

555
00:19:32,039 --> 00:19:37,080
around the world so Jaguar Ford General

556
00:19:34,260 --> 00:19:38,580
Motors all of these companies and the

557
00:19:37,080 --> 00:19:41,160
half of them that are subsidiaries of

558
00:19:38,580 --> 00:19:42,960
the other half and they collaborate on

559
00:19:41,160 --> 00:19:44,520
an open source Linux distro that is

560
00:19:42,960 --> 00:19:47,400
suitable for their needs generally

561
00:19:44,520 --> 00:19:49,799
across the entirety of what a car uses

562
00:19:47,400 --> 00:19:51,120
these things for it's not just the

563
00:19:49,799 --> 00:19:53,760
display that shows your cool Android

564
00:19:51,120 --> 00:19:55,080
games it's the the infotainment system

565
00:19:53,760 --> 00:19:57,000
for the rest of the car playing things

566
00:19:55,080 --> 00:19:58,200
in the back seat displays while the

567
00:19:57,000 --> 00:19:59,940
front seats can control them with

568
00:19:58,200 --> 00:20:01,679
parental controls it's having the

569
00:19:59,940 --> 00:20:04,020
connectivity for things like having the

570
00:20:01,679 --> 00:20:05,700
car with a built-in Wi-Fi hotspot it's

571
00:20:04,020 --> 00:20:07,980
for things like having the telematics so

572
00:20:05,700 --> 00:20:09,600
that your car can gather the data to go

573
00:20:07,980 --> 00:20:10,919
okay the tire pressure is low on the

574
00:20:09,600 --> 00:20:12,480
front right side I need to include a

575
00:20:10,919 --> 00:20:14,340
little icon in the top right corner of

576
00:20:12,480 --> 00:20:15,960
my lovely little display Dash rather

577
00:20:14,340 --> 00:20:17,940
than having an ugly light on the main

578
00:20:15,960 --> 00:20:20,580
section all of that data gets golden

579
00:20:17,940 --> 00:20:22,320
process through Linux based systems in

580
00:20:20,580 --> 00:20:24,120
some of these modern cars because the

581
00:20:22,320 --> 00:20:25,860
automotive grade Linux Foundation has

582
00:20:24,120 --> 00:20:27,179
actually stepped up and made one that

583
00:20:25,860 --> 00:20:29,160
they are happy to trust

584
00:20:27,179 --> 00:20:30,660
and they've done that by encouraging

585
00:20:29,160 --> 00:20:32,400
them to share their resources they've

586
00:20:30,660 --> 00:20:33,840
got these companies to go look

587
00:20:32,400 --> 00:20:35,460
we'll let you know what we're getting

588
00:20:33,840 --> 00:20:37,260
wrong and what's wrong we'll give you

589
00:20:35,460 --> 00:20:38,640
some money to even fix it and they've

590
00:20:37,260 --> 00:20:41,340
built this collaborative organization

591
00:20:38,640 --> 00:20:44,640
together with those companies to try and

592
00:20:41,340 --> 00:20:47,580
Foster a community first company is able

593
00:20:44,640 --> 00:20:49,980
to pull from open project and it's made

594
00:20:47,580 --> 00:20:51,539
a difference our cars are generally more

595
00:20:49,980 --> 00:20:52,799
secure now when they use this kind of a

596
00:20:51,539 --> 00:20:55,559
stack because people are actually

597
00:20:52,799 --> 00:20:58,440
checking it and that leads to a large

598
00:20:55,559 --> 00:21:01,799
part of what makes them actually a good

599
00:20:58,440 --> 00:21:04,320
example to pull from is that they're

600
00:21:01,799 --> 00:21:09,299
actually testing the damn stuff

601
00:21:04,320 --> 00:21:11,580
and that is no small fee because they

602
00:21:09,299 --> 00:21:13,440
build their own test Hardware because if

603
00:21:11,580 --> 00:21:15,299
you're testing a headless board in an

604
00:21:13,440 --> 00:21:17,520
embedded context and you might be doing

605
00:21:15,299 --> 00:21:19,440
over-the-air updates via a third-party

606
00:21:17,520 --> 00:21:21,600
wireless network provided to you by the

607
00:21:19,440 --> 00:21:23,700
car manufacturer whose pays and Telco

608
00:21:21,600 --> 00:21:25,440
you know a lump sum up front for every

609
00:21:23,700 --> 00:21:27,840
car that's solds like here's 15 grand

610
00:21:25,440 --> 00:21:30,179
for the next 50 years make that modem

611
00:21:27,840 --> 00:21:31,200
work so we can push updates which is the

612
00:21:30,179 --> 00:21:32,820
kind of thing they do in this

613
00:21:31,200 --> 00:21:34,080
environment you want to know that every

614
00:21:32,820 --> 00:21:35,460
time you push an update over that

615
00:21:34,080 --> 00:21:38,159
potentially unreliable cell phone

616
00:21:35,460 --> 00:21:40,500
network it will work which means that

617
00:21:38,159 --> 00:21:42,840
you have to have silent always not going

618
00:21:40,500 --> 00:21:44,520
to fail on you a b updates working on a

619
00:21:42,840 --> 00:21:46,080
headless environment you have no ability

620
00:21:44,520 --> 00:21:48,059
to go in and inspect

621
00:21:46,080 --> 00:21:51,120
and they test for that they build test

622
00:21:48,059 --> 00:21:53,280
Hardware like those SD mux cards and the

623
00:21:51,120 --> 00:21:56,220
SD wire board that you see on the on the

624
00:21:53,280 --> 00:21:58,320
right hand side there and that lets them

625
00:21:56,220 --> 00:22:00,360
do device under test headless boot

626
00:21:58,320 --> 00:22:02,460
testing and other low-level Hardware

627
00:22:00,360 --> 00:22:05,400
testing to guarantee that their Hardware

628
00:22:02,460 --> 00:22:07,980
that that software is designed for that

629
00:22:05,400 --> 00:22:11,220
they say it's good enough for a car to

630
00:22:07,980 --> 00:22:12,780
rely on for its safety features

631
00:22:11,220 --> 00:22:15,240
they're putting the hard work into

632
00:22:12,780 --> 00:22:17,960
testing that and that is absolutely

633
00:22:15,240 --> 00:22:21,240
critical to have that kind of exhaustive

634
00:22:17,960 --> 00:22:24,480
expensive labor-intensive

635
00:22:21,240 --> 00:22:26,880
absolutely necessary testing done for it

636
00:22:24,480 --> 00:22:29,039
to make a difference in the Aerospace

637
00:22:26,880 --> 00:22:30,299
context if your drone is going to fall

638
00:22:29,039 --> 00:22:32,159
out of the sky because the battery is

639
00:22:30,299 --> 00:22:33,480
low and it lags and you don't know the

640
00:22:32,159 --> 00:22:34,740
battery is low

641
00:22:33,480 --> 00:22:36,659
well you don't want that to fall on

642
00:22:34,740 --> 00:22:38,400
someone's head do you you want that to

643
00:22:36,659 --> 00:22:39,179
land safely and it won't be able to do

644
00:22:38,400 --> 00:22:41,100
that

645
00:22:39,179 --> 00:22:43,799
if it's not tested properly it can't

646
00:22:41,100 --> 00:22:46,080
miss signals it needs to never miss the

647
00:22:43,799 --> 00:22:47,640
signals that are important and unless

648
00:22:46,080 --> 00:22:48,900
the testing gets done you can't be sure

649
00:22:47,640 --> 00:22:50,760
of that which means you can't get a

650
00:22:48,900 --> 00:22:53,220
safety certification

651
00:22:50,760 --> 00:22:55,200
and then you can't sell which is why the

652
00:22:53,220 --> 00:22:56,640
companies care and why the companies put

653
00:22:55,200 --> 00:22:58,020
money into the project and why the

654
00:22:56,640 --> 00:22:59,760
project can afford to design their own

655
00:22:58,020 --> 00:23:02,760
test Hardware like this

656
00:22:59,760 --> 00:23:04,020
and that matters it affords their CI

657
00:23:02,760 --> 00:23:05,340
infrastructure a large part of the there

658
00:23:04,020 --> 00:23:08,039
was a talk earlier today about the

659
00:23:05,340 --> 00:23:09,900
kernel CI Pipeline and a contributor to

660
00:23:08,039 --> 00:23:12,299
that effort has been the automotive

661
00:23:09,900 --> 00:23:14,159
grade Linux project because they care

662
00:23:12,299 --> 00:23:17,460
about making sure that the kernel is CI

663
00:23:14,159 --> 00:23:19,799
tested all the damn time because unless

664
00:23:17,460 --> 00:23:21,659
it's tested all the damn time

665
00:23:19,799 --> 00:23:24,059
One update goes wrong

666
00:23:21,659 --> 00:23:27,179
one firmware rollout goes wrong

667
00:23:24,059 --> 00:23:29,940
one safety critical system goes wrong

668
00:23:27,179 --> 00:23:32,280
and someone may if it's a good day wind

669
00:23:29,940 --> 00:23:35,039
up in a ditch if it's a bad day wind up

670
00:23:32,280 --> 00:23:37,440
off a cliff and the car companies don't

671
00:23:35,039 --> 00:23:38,220
want that so they pay and they make it

672
00:23:37,440 --> 00:23:40,919
work

673
00:23:38,220 --> 00:23:42,480
but the community had to move first

674
00:23:40,919 --> 00:23:44,400
which is why I think we need to start

675
00:23:42,480 --> 00:23:45,600
moving in this direction for Aerospace

676
00:23:44,400 --> 00:23:48,179
software we need to actually begin

677
00:23:45,600 --> 00:23:50,400
building a community group like the

678
00:23:48,179 --> 00:23:52,860
automotive grade Linux Foundation

679
00:23:50,400 --> 00:23:54,179
for Aerospace because we have this world

680
00:23:52,860 --> 00:23:56,280
where drones are being flown around with

681
00:23:54,179 --> 00:23:58,380
Linux on board we've got multiple

682
00:23:56,280 --> 00:24:00,419
systems coordinating cameras and flight

683
00:23:58,380 --> 00:24:01,919
software on Linux powered drones that

684
00:24:00,419 --> 00:24:03,539
hobbyists are doing certainly but

685
00:24:01,919 --> 00:24:05,760
probably the commercial companies too at

686
00:24:03,539 --> 00:24:07,559
this point because well if the code is

687
00:24:05,760 --> 00:24:08,880
up there and I want to make my drone for

688
00:24:07,559 --> 00:24:10,140
under a hundred dollars how am I going

689
00:24:08,880 --> 00:24:11,760
to do that as easily as possible I'll

690
00:24:10,140 --> 00:24:14,659
just go yoink

691
00:24:11,760 --> 00:24:17,220
and it's probably happening yesterday

692
00:24:14,659 --> 00:24:19,380
and that means that the risk is already

693
00:24:17,220 --> 00:24:21,720
out there in a

694
00:24:19,380 --> 00:24:23,159
strict sense but not a practical sense

695
00:24:21,720 --> 00:24:24,900
because the companies don't have any

696
00:24:23,159 --> 00:24:26,580
liability right now there is no openness

697
00:24:24,900 --> 00:24:28,559
they aren't talking about how they've

698
00:24:26,580 --> 00:24:30,299
built these things internally because

699
00:24:28,559 --> 00:24:32,280
there's no incentive for them to do so

700
00:24:30,299 --> 00:24:34,620
all of the things I was saying before

701
00:24:32,280 --> 00:24:36,720
are why no one says that these things

702
00:24:34,620 --> 00:24:39,059
are being used this way there is a

703
00:24:36,720 --> 00:24:41,340
reason why I have not been able to find

704
00:24:39,059 --> 00:24:44,159
a single pull request from SpaceX it's

705
00:24:41,340 --> 00:24:45,659
because when they used Linux on the

706
00:24:44,159 --> 00:24:48,059
starlink satellites on the dragon

707
00:24:45,659 --> 00:24:50,460
spacecraft there is liability for them

708
00:24:48,059 --> 00:24:52,860
massive commercial money on the table

709
00:24:50,460 --> 00:24:54,539
liability lives on the line important

710
00:24:52,860 --> 00:24:56,159
liability that they pay a team of

711
00:24:54,539 --> 00:24:57,780
lawyers to look after

712
00:24:56,159 --> 00:24:58,620
and if they ever publish what they're

713
00:24:57,780 --> 00:25:00,539
using

714
00:24:58,620 --> 00:25:02,460
that gets worse for them

715
00:25:00,539 --> 00:25:04,440
because they're the ones on the hook for

716
00:25:02,460 --> 00:25:06,600
it not the people who originally made it

717
00:25:04,440 --> 00:25:08,400
not the people who were in a position to

718
00:25:06,600 --> 00:25:10,799
publish it openly in a collaborative way

719
00:25:08,400 --> 00:25:13,020
under those kinds of itar restrictions

720
00:25:10,799 --> 00:25:14,580
that would be all their burden and they

721
00:25:13,020 --> 00:25:15,780
don't want that burden which is why

722
00:25:14,580 --> 00:25:18,120
you've never seen a pull request from

723
00:25:15,780 --> 00:25:19,799
SpaceX and why we can do something about

724
00:25:18,120 --> 00:25:21,960
changing that if we try and set up an

725
00:25:19,799 --> 00:25:24,600
organization to Steward this kind of

726
00:25:21,960 --> 00:25:27,480
work and try and create a lack of a

727
00:25:24,600 --> 00:25:29,279
better name for it and Aerospace Linux

728
00:25:27,480 --> 00:25:30,779
group

729
00:25:29,279 --> 00:25:32,039
probably have to find a better name

730
00:25:30,779 --> 00:25:33,299
because automotive and Aerospace same

731
00:25:32,039 --> 00:25:35,340
acronym but

732
00:25:33,299 --> 00:25:37,340
something like that is needed if we want

733
00:25:35,340 --> 00:25:39,840
to try and move this forward because

734
00:25:37,340 --> 00:25:42,900
Starship is going to have Linux on it

735
00:25:39,840 --> 00:25:44,520
SpaceX use Linux everywhere

736
00:25:42,900 --> 00:25:46,320
every time they launched a batch of

737
00:25:44,520 --> 00:25:48,179
their styling satellites back when they

738
00:25:46,320 --> 00:25:50,279
were launching them 60 at a time

739
00:25:48,179 --> 00:25:52,140
they were launching 4 000 Linux

740
00:25:50,279 --> 00:25:55,320
computers into space they were literally

741
00:25:52,140 --> 00:25:57,360
launching 75 separate Linux computers on

742
00:25:55,320 --> 00:25:59,700
every one of those satellites

743
00:25:57,360 --> 00:26:02,460
75 computers in a tightly coupled

744
00:25:59,700 --> 00:26:04,740
supervisory control domain dense

745
00:26:02,460 --> 00:26:06,360
embedded network doing millions of

746
00:26:04,740 --> 00:26:08,700
things all the time flying around up

747
00:26:06,360 --> 00:26:10,380
there with engines that could slam them

748
00:26:08,700 --> 00:26:12,299
into other people's satellites or God

749
00:26:10,380 --> 00:26:14,340
Knows Why

750
00:26:12,299 --> 00:26:16,380
and we have no idea what's going on with

751
00:26:14,340 --> 00:26:18,240
it we just have to accept that they've

752
00:26:16,380 --> 00:26:20,159
done their hard work and that's for the

753
00:26:18,240 --> 00:26:22,320
most part fine

754
00:26:20,159 --> 00:26:24,059
but should forever remain that way do we

755
00:26:22,320 --> 00:26:25,799
want it to remain proprietary do we want

756
00:26:24,059 --> 00:26:27,539
it to remain behind closed doors like

757
00:26:25,799 --> 00:26:29,640
that or do we want to actually try and

758
00:26:27,539 --> 00:26:30,960
push the industry forward like we did in

759
00:26:29,640 --> 00:26:33,539
Automotive

760
00:26:30,960 --> 00:26:34,919
and one of the ways to do that is to

761
00:26:33,539 --> 00:26:36,720
start building that organization start

762
00:26:34,919 --> 00:26:39,059
building that group and start working on

763
00:26:36,720 --> 00:26:41,159
the important projects to make that more

764
00:26:39,059 --> 00:26:42,720
valuable for the companies so that they

765
00:26:41,159 --> 00:26:44,279
might choose to adopt it because that is

766
00:26:42,720 --> 00:26:46,380
what the Automotive Group did they

767
00:26:44,279 --> 00:26:48,600
spearheaded tizen that's actually where

768
00:26:46,380 --> 00:26:50,039
it came from first not the phones it

769
00:26:48,600 --> 00:26:52,500
actually came out as part of this this

770
00:26:50,039 --> 00:26:54,360
Core group and ties in one of those

771
00:26:52,500 --> 00:26:56,940
first wins was Automotive for the

772
00:26:54,360 --> 00:26:59,159
display and the infotainment system and

773
00:26:56,940 --> 00:27:00,659
then they expanded it they built a lot

774
00:26:59,159 --> 00:27:03,000
of work on top of it to make it reliable

775
00:27:00,659 --> 00:27:04,980
and robust and they did that testing

776
00:27:03,000 --> 00:27:06,779
work and they've built it up so that it

777
00:27:04,980 --> 00:27:07,799
is able to do the infotainment the

778
00:27:06,779 --> 00:27:09,600
telematics all of this stuff

779
00:27:07,799 --> 00:27:11,700
simultaneously to a level where they are

780
00:27:09,600 --> 00:27:13,020
happy that it's good enough and we have

781
00:27:11,700 --> 00:27:15,299
to do the same thing for Aerospace

782
00:27:13,020 --> 00:27:17,460
software or else the companies can't use

783
00:27:15,299 --> 00:27:19,320
it and start the cycle because they're

784
00:27:17,460 --> 00:27:20,760
not going to take their old code it was

785
00:27:19,320 --> 00:27:22,500
a footnote in one of the earlier slides

786
00:27:20,760 --> 00:27:23,760
but I don't think they're ever going to

787
00:27:22,500 --> 00:27:25,740
open source what they've written right

788
00:27:23,760 --> 00:27:27,600
now because if they were to open source

789
00:27:25,740 --> 00:27:29,580
what they've written right now that

790
00:27:27,600 --> 00:27:32,400
would be a risk to them that they cannot

791
00:27:29,580 --> 00:27:34,440
stomach if they took that software that

792
00:27:32,400 --> 00:27:35,940
was already flying and let other people

793
00:27:34,440 --> 00:27:37,919
see it

794
00:27:35,940 --> 00:27:38,940
you look back at something like the 737

795
00:27:37,919 --> 00:27:40,919
Max

796
00:27:38,940 --> 00:27:42,900
that was a software problem

797
00:27:40,919 --> 00:27:44,159
100 a software problem there was some

798
00:27:42,900 --> 00:27:45,779
Hardware involved but it was a software

799
00:27:44,159 --> 00:27:47,640
problem that killed people

800
00:27:45,779 --> 00:27:49,080
quite a lot of people for a plane crash

801
00:27:47,640 --> 00:27:50,340
kind of thing actually

802
00:27:49,080 --> 00:27:53,460
and

803
00:27:50,340 --> 00:27:55,640
if Boeing had been open about that

804
00:27:53,460 --> 00:27:59,100
software they may have been

805
00:27:55,640 --> 00:28:02,039
worse out there there may still be worse

806
00:27:59,100 --> 00:28:04,679
in that code and we don't know they

807
00:28:02,039 --> 00:28:06,720
probably have some idea but calculated

808
00:28:04,679 --> 00:28:08,100
risk management is what it is for large

809
00:28:06,720 --> 00:28:10,320
companies like that

810
00:28:08,100 --> 00:28:12,240
and they will never want to open source

811
00:28:10,320 --> 00:28:14,460
that kind of code because that is too

812
00:28:12,240 --> 00:28:16,380
big of a risk for them

813
00:28:14,460 --> 00:28:18,120
and that is just how business Works they

814
00:28:16,380 --> 00:28:19,740
have to stay solvent or else well they

815
00:28:18,120 --> 00:28:21,179
stop being a business so they won't open

816
00:28:19,740 --> 00:28:23,700
source that software they will

817
00:28:21,179 --> 00:28:25,860
potentially adopt new code new code that

818
00:28:23,700 --> 00:28:27,120
has been proven good enough

819
00:28:25,860 --> 00:28:28,140
but we're not going to get a big code

820
00:28:27,120 --> 00:28:30,120
dump we're not going to get someone

821
00:28:28,140 --> 00:28:32,340
going here is the last 30 years of hard

822
00:28:30,120 --> 00:28:35,159
work to make amazing thing because

823
00:28:32,340 --> 00:28:36,659
fundamentally it is like deciding to run

824
00:28:35,159 --> 00:28:39,179
through a plate glass window for these

825
00:28:36,659 --> 00:28:40,799
companies that is utterly insane levels

826
00:28:39,179 --> 00:28:42,480
of risk and they will not take it on the

827
00:28:40,799 --> 00:28:44,039
community has to move first and we have

828
00:28:42,480 --> 00:28:45,840
to start doing things like having

829
00:28:44,039 --> 00:28:49,020
significant testing on the real-time

830
00:28:45,840 --> 00:28:50,340
preempt kernel set it's almost all the

831
00:28:49,020 --> 00:28:52,500
way in now there's just a few little

832
00:28:50,340 --> 00:28:54,840
extra patches left and that's a big win

833
00:28:52,500 --> 00:28:57,299
over the last 12 months 12 months ago

834
00:28:54,840 --> 00:28:59,640
there was still mostly unmerged but

835
00:28:57,299 --> 00:29:01,919
ready for merge now it's merged mostly

836
00:28:59,640 --> 00:29:03,600
and just a little bit living out that is

837
00:29:01,919 --> 00:29:06,000
a big win because a lot of Aerospace

838
00:29:03,600 --> 00:29:07,860
stuff needs real-time systems it needs

839
00:29:06,000 --> 00:29:09,900
extremely precise timing response it

840
00:29:07,860 --> 00:29:12,840
needs to acknowledge updates in a very

841
00:29:09,900 --> 00:29:15,419
important Fashion on hierarchy you need

842
00:29:12,840 --> 00:29:17,220
to have I must know now and I need to

843
00:29:15,419 --> 00:29:18,480
know really soon and everything else can

844
00:29:17,220 --> 00:29:19,919
wait after that

845
00:29:18,480 --> 00:29:22,020
and unless you have those kind of

846
00:29:19,919 --> 00:29:24,419
patches you can't get that guarantee in

847
00:29:22,020 --> 00:29:26,520
the Linux kernel it's just how it works

848
00:29:24,419 --> 00:29:27,720
and so that's a crucial component that's

849
00:29:26,520 --> 00:29:29,640
the kind of thing that could be tested

850
00:29:27,720 --> 00:29:31,500
more exhaustively there are some other

851
00:29:29,640 --> 00:29:33,299
open source projects that are in this

852
00:29:31,500 --> 00:29:35,039
Aerospace domain that could definitely

853
00:29:33,299 --> 00:29:36,779
deserve more Community attention from

854
00:29:35,039 --> 00:29:39,299
people who aren't already working at

855
00:29:36,779 --> 00:29:41,100
large companies there's qbos which tried

856
00:29:39,299 --> 00:29:42,960
to make an open source operating system

857
00:29:41,100 --> 00:29:44,700
for satellites you can actually build

858
00:29:42,960 --> 00:29:45,899
that one still on a Raspberry Pi it's a

859
00:29:44,700 --> 00:29:48,000
great little project but the company

860
00:29:45,899 --> 00:29:49,260
shifted away from it because they

861
00:29:48,000 --> 00:29:51,059
weren't making much money off that

862
00:29:49,260 --> 00:29:52,679
compared to how much development time it

863
00:29:51,059 --> 00:29:54,179
took for them to do the kind of testing

864
00:29:52,679 --> 00:29:56,100
and work that was needed to make it

865
00:29:54,179 --> 00:29:57,600
useful they shifted into running a

866
00:29:56,100 --> 00:29:59,640
management software stack that they had

867
00:29:57,600 --> 00:30:01,440
built side by side with it so kind of a

868
00:29:59,640 --> 00:30:04,860
grafana for your satellite kind of a

869
00:30:01,440 --> 00:30:06,600
thing and that's fine they've opened

870
00:30:04,860 --> 00:30:09,299
that code which is great it's there as a

871
00:30:06,600 --> 00:30:11,159
good example but it's not one people can

872
00:30:09,299 --> 00:30:12,480
just pick up and use right now it has

873
00:30:11,159 --> 00:30:14,159
missing pieces that were previously

874
00:30:12,480 --> 00:30:15,779
proprietary to do with radios and all

875
00:30:14,159 --> 00:30:17,520
sorts of other components but it could

876
00:30:15,779 --> 00:30:19,620
be a starting point if a project wants

877
00:30:17,520 --> 00:30:20,820
to pick it up and then there's NASA's

878
00:30:19,620 --> 00:30:22,440
core flight system that's actually

879
00:30:20,820 --> 00:30:24,240
running real satellites

880
00:30:22,440 --> 00:30:26,039
but once you learn how that patch

881
00:30:24,240 --> 00:30:28,860
process works and how that review cycle

882
00:30:26,039 --> 00:30:31,260
goes you really do get a very quick

883
00:30:28,860 --> 00:30:32,640
lesson in how Aerospace software works

884
00:30:31,260 --> 00:30:34,919
if you want to contribute to that

885
00:30:32,640 --> 00:30:36,779
project you're basically proposing a

886
00:30:34,919 --> 00:30:38,700
contribution not actually giving them

887
00:30:36,779 --> 00:30:40,080
one they may rewrite it completely

888
00:30:38,700 --> 00:30:41,880
because

889
00:30:40,080 --> 00:30:43,799
if you're from a certain country or from

890
00:30:41,880 --> 00:30:45,779
a certain place or a certain there's a

891
00:30:43,799 --> 00:30:48,000
million reasons why they may completely

892
00:30:45,779 --> 00:30:49,620
rewrite it based on your suggestion

893
00:30:48,000 --> 00:30:52,200
or they just may review it and

894
00:30:49,620 --> 00:30:54,539
potentially include it it's a heavily

895
00:30:52,200 --> 00:30:56,760
gated process because well NASA actually

896
00:30:54,539 --> 00:30:59,760
used that to fly satellites so they're

897
00:30:56,760 --> 00:31:01,440
damn careful with it and rightly so but

898
00:30:59,760 --> 00:31:03,779
there is a good bootstrapping project

899
00:31:01,440 --> 00:31:05,580
the open sack kit that gets that all

900
00:31:03,779 --> 00:31:07,260
running up on a Linux you know Linux

901
00:31:05,580 --> 00:31:09,059
environment and a virtual machine and

902
00:31:07,260 --> 00:31:10,860
lets you play with it but Linux isn't a

903
00:31:09,059 --> 00:31:13,200
first class platform for them

904
00:31:10,860 --> 00:31:14,039
because it's not reliable enough in

905
00:31:13,200 --> 00:31:15,539
their View

906
00:31:14,039 --> 00:31:16,919
well we could try and make it more

907
00:31:15,539 --> 00:31:18,240
reliable we could try and push that

908
00:31:16,919 --> 00:31:19,440
forward to get it to the point where

909
00:31:18,240 --> 00:31:21,659
they considered it a first class

910
00:31:19,440 --> 00:31:23,820
platform and not just a good enough for

911
00:31:21,659 --> 00:31:25,799
a VM on your desktop

912
00:31:23,820 --> 00:31:29,000
there's F Prime that's actually flying a

913
00:31:25,799 --> 00:31:29,000
helicopter on Mars we

914
00:31:31,919 --> 00:31:35,940
are openly on regular stuff on Earth we

915
00:31:34,440 --> 00:31:38,460
could see more drones being developed

916
00:31:35,940 --> 00:31:39,480
with f Prime in an open way not in a

917
00:31:38,460 --> 00:31:40,799
closed way because I wouldn't be

918
00:31:39,480 --> 00:31:43,080
surprised to see a lot of commercial

919
00:31:40,799 --> 00:31:44,399
drones using F Prime if they ever told

920
00:31:43,080 --> 00:31:47,340
us what they used

921
00:31:44,399 --> 00:31:49,020
and absolutely critical is maintaining

922
00:31:47,340 --> 00:31:51,059
that real-time preempt Aerospace

923
00:31:49,020 --> 00:31:53,460
software has a very strong need for real

924
00:31:51,059 --> 00:31:54,659
time I cannot emphasize this enough when

925
00:31:53,460 --> 00:31:56,340
you have to move out of the way of

926
00:31:54,659 --> 00:31:57,000
something while moving very quickly at

927
00:31:56,340 --> 00:31:59,039
like

928
00:31:57,000 --> 00:32:02,399
hundreds of kilometers an hour or even

929
00:31:59,039 --> 00:32:04,620
worse thousands of kilometers a second

930
00:32:02,399 --> 00:32:06,059
you need to know about it right now

931
00:32:04,620 --> 00:32:07,020
otherwise you could be in serious

932
00:32:06,059 --> 00:32:07,919
trouble

933
00:32:07,020 --> 00:32:09,360
and

934
00:32:07,919 --> 00:32:11,880
really that is where the bigger picture

935
00:32:09,360 --> 00:32:14,039
is it is testing it is having a cohesive

936
00:32:11,880 --> 00:32:16,140
group to manage that testing to deal

937
00:32:14,039 --> 00:32:18,539
with the funding of that testing and to

938
00:32:16,140 --> 00:32:20,460
Steward the group as a whole like how we

939
00:32:18,539 --> 00:32:22,740
have already done with automotive and I

940
00:32:20,460 --> 00:32:25,919
think we can push this forward it just

941
00:32:22,740 --> 00:32:27,899
needs to start and unfortunately for me

942
00:32:25,919 --> 00:32:30,600
I'm in the same position as some of

943
00:32:27,899 --> 00:32:32,580
those companies so I can't start it I

944
00:32:30,600 --> 00:32:35,100
need to try and encourage the community

945
00:32:32,580 --> 00:32:38,159
to start so I can join

946
00:32:35,100 --> 00:32:40,760
so with that and 10 minutes left I'll

947
00:32:38,159 --> 00:32:40,760
take questions

948
00:32:50,760 --> 00:32:55,440
so did you actually end up building a

949
00:32:52,559 --> 00:32:58,399
satellite oh yeah but I can't fly it

950
00:32:55,440 --> 00:32:58,399
works on my desk

951
00:33:01,200 --> 00:33:05,340
since we already have Automotive grade

952
00:33:02,940 --> 00:33:07,500
Linux could the scope of that be

953
00:33:05,340 --> 00:33:09,000
expanded to cover Aerospace or are there

954
00:33:07,500 --> 00:33:10,380
do they need to be maintained as two

955
00:33:09,000 --> 00:33:12,720
separate and largely overlapping

956
00:33:10,380 --> 00:33:14,039
projects it could but I don't know how

957
00:33:12,720 --> 00:33:16,380
well that would help the organization

958
00:33:14,039 --> 00:33:18,179
for their needs they do have a very

959
00:33:16,380 --> 00:33:20,640
different set of needs and while there

960
00:33:18,179 --> 00:33:22,380
is significant overlap at the core I

961
00:33:20,640 --> 00:33:24,120
don't know how much benefit there would

962
00:33:22,380 --> 00:33:26,820
be in the long run to those peripheral

963
00:33:24,120 --> 00:33:29,460
interests the whole infotainment portion

964
00:33:26,820 --> 00:33:31,500
is a great clear-cut example there you

965
00:33:29,460 --> 00:33:33,480
know all the Aerospace stuff broadly

966
00:33:31,500 --> 00:33:35,220
speaking outside of very specific niches

967
00:33:33,480 --> 00:33:36,779
like large airliners where you might

968
00:33:35,220 --> 00:33:39,360
have a whole infotainment system in them

969
00:33:36,779 --> 00:33:41,279
outside of that there isn't really a

970
00:33:39,360 --> 00:33:42,899
need for large chunks of what those

971
00:33:41,279 --> 00:33:45,120
people are building and testing and

972
00:33:42,899 --> 00:33:47,159
making sure it works so

973
00:33:45,120 --> 00:33:48,539
potentially it really does come down to

974
00:33:47,159 --> 00:33:50,399
how much is done better as a

975
00:33:48,539 --> 00:33:52,260
collaboration between two peer

976
00:33:50,399 --> 00:33:54,059
organizations or whether it would be

977
00:33:52,260 --> 00:33:55,380
better to make it part of their overall

978
00:33:54,059 --> 00:33:57,480
mission and whether they want to adopt

979
00:33:55,380 --> 00:33:59,279
that they are funded at the moment by

980
00:33:57,480 --> 00:34:01,080
majority car companies

981
00:33:59,279 --> 00:34:03,779
those car companies may not want to have

982
00:34:01,080 --> 00:34:05,279
other people you know butting in

983
00:34:03,779 --> 00:34:06,960
they want that control potentially

984
00:34:05,279 --> 00:34:08,099
because that serves their needs they're

985
00:34:06,960 --> 00:34:12,919
paying for that

986
00:34:08,099 --> 00:34:12,919
so it may be but potentially no

987
00:34:16,139 --> 00:34:18,740
any more questions

988
00:34:27,300 --> 00:34:34,080
you said if I remember correctly that

989
00:34:29,899 --> 00:34:36,659
NASA didn't trust Unix for Linux for uh

990
00:34:34,080 --> 00:34:39,720
most things yet apparently they trusted

991
00:34:36,659 --> 00:34:42,480
it for helicopter on Mars how did that

992
00:34:39,720 --> 00:34:44,940
that's got an interesting story actually

993
00:34:42,480 --> 00:34:48,240
um in in that particular case it kind of

994
00:34:44,940 --> 00:34:50,520
came down to they were forced to

995
00:34:48,240 --> 00:34:52,800
um the the better example is that they

996
00:34:50,520 --> 00:34:54,300
were willing to trust the cameras on the

997
00:34:52,800 --> 00:34:55,700
on the landing portion of that mission

998
00:34:54,300 --> 00:34:58,740
to Linux

999
00:34:55,700 --> 00:35:00,839
because with the helicopter they were

1000
00:34:58,740 --> 00:35:02,220
forced to take that one because the only

1001
00:35:00,839 --> 00:35:04,020
way they could get enough compute power

1002
00:35:02,220 --> 00:35:06,240
to do that mission to have that

1003
00:35:04,020 --> 00:35:08,339
helicopter fly keep that under the

1004
00:35:06,240 --> 00:35:10,200
weight threshold to make these able to

1005
00:35:08,339 --> 00:35:12,660
lift off with the very tight power

1006
00:35:10,200 --> 00:35:14,280
budget they had that the thing is right

1007
00:35:12,660 --> 00:35:15,480
on the razor's Edge in terms of how much

1008
00:35:14,280 --> 00:35:16,800
weight they were able to get on there

1009
00:35:15,480 --> 00:35:18,200
and how much power they were able to

1010
00:35:16,800 --> 00:35:21,420
store and how much they needed to fly

1011
00:35:18,200 --> 00:35:23,099
was to run with a Linux board and their

1012
00:35:21,420 --> 00:35:26,480
decision was based on okay we can either

1013
00:35:23,099 --> 00:35:29,220
try and port a whole thing over to this

1014
00:35:26,480 --> 00:35:31,260
or we can take what's already on there

1015
00:35:29,220 --> 00:35:33,839
that's good enough that was in fact

1016
00:35:31,260 --> 00:35:35,280
actually supported by the the air the

1017
00:35:33,839 --> 00:35:37,680
automotive grade Linux so they were able

1018
00:35:35,280 --> 00:35:39,720
to use a quite robust hardened kernel

1019
00:35:37,680 --> 00:35:41,520
tested line of code it was all custom

1020
00:35:39,720 --> 00:35:44,700
down but they could start from somewhere

1021
00:35:41,520 --> 00:35:46,859
that trusted and tested they were able

1022
00:35:44,700 --> 00:35:48,780
to take that and still build a mission

1023
00:35:46,859 --> 00:35:49,980
and then on the other side of it on the

1024
00:35:48,780 --> 00:35:51,900
receive side there's actually another

1025
00:35:49,980 --> 00:35:53,460
little computer on the Rover that

1026
00:35:51,900 --> 00:35:55,500
listens to the helicopter that also runs

1027
00:35:53,460 --> 00:35:57,599
Linux just mainly because it was

1028
00:35:55,500 --> 00:36:00,900
convenient for the team to do it all on

1029
00:35:57,599 --> 00:36:03,420
Linux as one stack and that one actually

1030
00:36:00,900 --> 00:36:05,760
talks via a uart to the main Rover

1031
00:36:03,420 --> 00:36:07,619
operating system and it's not like that

1032
00:36:05,760 --> 00:36:09,420
actually gets to talk back to earth when

1033
00:36:07,619 --> 00:36:11,099
they get files from the helicopter when

1034
00:36:09,420 --> 00:36:12,480
it's done with a flight it just

1035
00:36:11,099 --> 00:36:14,460
literally writes them out to a Serial

1036
00:36:12,480 --> 00:36:15,900
buffer and then if the Rover is happy

1037
00:36:14,460 --> 00:36:17,760
with that it will then write that into

1038
00:36:15,900 --> 00:36:20,280
memory and later do its thing with it

1039
00:36:17,760 --> 00:36:21,359
there is layers and gatekeeping in this

1040
00:36:20,280 --> 00:36:23,880
process

1041
00:36:21,359 --> 00:36:26,520
that is part of their willingness to

1042
00:36:23,880 --> 00:36:29,040
accept Linux in this context is because

1043
00:36:26,520 --> 00:36:31,619
it's behind safety's not in critical

1044
00:36:29,040 --> 00:36:33,720
places was the only option to get it

1045
00:36:31,619 --> 00:36:35,400
done on a deadline they were kind of

1046
00:36:33,720 --> 00:36:38,099
forced into it and willing to accept

1047
00:36:35,400 --> 00:36:40,020
that under certain key circumstances and

1048
00:36:38,099 --> 00:36:41,820
that's not to say it was a bad choice

1049
00:36:40,020 --> 00:36:43,500
it's a good choice it flies that thing

1050
00:36:41,820 --> 00:36:46,920
well

1051
00:36:43,500 --> 00:36:48,780
but under other circumstances such as

1052
00:36:46,920 --> 00:36:49,920
for instance one of the new missions

1053
00:36:48,780 --> 00:36:51,359
that's currently under development

1054
00:36:49,920 --> 00:36:53,339
that's going to do a large I think it's

1055
00:36:51,359 --> 00:36:55,020
a hexacopter I might be remembering that

1056
00:36:53,339 --> 00:36:57,780
wrong but a very large multi-rotor

1057
00:36:55,020 --> 00:37:00,540
copter being sent to Titan

1058
00:36:57,780 --> 00:37:01,800
that probably won't run Linux that's

1059
00:37:00,540 --> 00:37:03,240
probably going to be running a real-time

1060
00:37:01,800 --> 00:37:04,560
operating system and a lot of custom

1061
00:37:03,240 --> 00:37:06,839
stuff on top

1062
00:37:04,560 --> 00:37:08,160
because they can't risk it this is going

1063
00:37:06,839 --> 00:37:09,960
to be a nuclear-powered flying

1064
00:37:08,160 --> 00:37:13,260
helicopter on you know Saturn's moons

1065
00:37:09,960 --> 00:37:15,960
they don't want anything to go wrong

1066
00:37:13,260 --> 00:37:17,220
and they won't trust Linux there and

1067
00:37:15,960 --> 00:37:20,700
this is the thing that needs to try and

1068
00:37:17,220 --> 00:37:23,220
change but with the Imaging cameras that

1069
00:37:20,700 --> 00:37:24,900
was an interesting one they couldn't get

1070
00:37:23,220 --> 00:37:27,900
all of that to work without using a

1071
00:37:24,900 --> 00:37:30,240
regular computer they had too much data

1072
00:37:27,900 --> 00:37:32,520
so the Imaging cameras the the wonderful

1073
00:37:30,240 --> 00:37:34,320
videos that we have this time as opposed

1074
00:37:32,520 --> 00:37:36,480
to last time when Ingenuity was Landing

1075
00:37:34,320 --> 00:37:38,640
this time was because they actually took

1076
00:37:36,480 --> 00:37:40,380
an industrial control computer that was

1077
00:37:38,640 --> 00:37:41,640
used for managing you know regular

1078
00:37:40,380 --> 00:37:43,680
production lines one of these little PC

1079
00:37:41,640 --> 00:37:45,540
104 kind of embedded boards but a modern

1080
00:37:43,680 --> 00:37:46,980
one with lots of bandwidth

1081
00:37:45,540 --> 00:37:48,300
made sure it was good enough for the job

1082
00:37:46,980 --> 00:37:49,260
the G's and the red hardening and

1083
00:37:48,300 --> 00:37:51,780
everything

1084
00:37:49,260 --> 00:37:53,280
and then they plugged a ton of USB

1085
00:37:51,780 --> 00:37:55,500
cameras into it

1086
00:37:53,280 --> 00:37:56,880
so all of those wonderful multi-views of

1087
00:37:55,500 --> 00:37:58,020
it as it's coming down and all the

1088
00:37:56,880 --> 00:38:00,000
photos that we get from where it's

1089
00:37:58,020 --> 00:38:01,800
dropped things and stuff's coming behind

1090
00:38:00,000 --> 00:38:03,240
it and all of the wonderful multi-camera

1091
00:38:01,800 --> 00:38:04,079
reviews we have of that seven minutes of

1092
00:38:03,240 --> 00:38:06,240
Terror

1093
00:38:04,079 --> 00:38:08,460
they came in from one single Linux box

1094
00:38:06,240 --> 00:38:09,480
because it was the absolute right tool

1095
00:38:08,460 --> 00:38:11,099
for the job

1096
00:38:09,480 --> 00:38:12,359
because if they tried to do that with a

1097
00:38:11,099 --> 00:38:13,800
real-time one they would have had to

1098
00:38:12,359 --> 00:38:15,900
basically build half of Linux from

1099
00:38:13,800 --> 00:38:17,820
scratch to do all of that for themselves

1100
00:38:15,900 --> 00:38:20,579
they had to send that data they had to

1101
00:38:17,820 --> 00:38:22,560
do jpeg processing to compress it down

1102
00:38:20,579 --> 00:38:24,300
so they could send it back they had to

1103
00:38:22,560 --> 00:38:25,920
manage the multiple i o streams they had

1104
00:38:24,300 --> 00:38:28,140
to do USB controlling from multiple

1105
00:38:25,920 --> 00:38:30,780
points and you know all of that stuff in

1106
00:38:28,140 --> 00:38:33,540
an operating system a full one that does

1107
00:38:30,780 --> 00:38:35,040
lots of stuff on desktops and even more

1108
00:38:33,540 --> 00:38:37,320
stuff on servers

1109
00:38:35,040 --> 00:38:39,420
it already does that and that was the

1110
00:38:37,320 --> 00:38:41,640
solution that was right for that job

1111
00:38:39,420 --> 00:38:43,320
that's a good example of where Linux was

1112
00:38:41,640 --> 00:38:44,760
the right tool and they willingly chose

1113
00:38:43,320 --> 00:38:46,200
it

1114
00:38:44,760 --> 00:38:47,940
but it doesn't get as much publicity

1115
00:38:46,200 --> 00:38:50,160
because it's not a cool little Rover or

1116
00:38:47,940 --> 00:38:52,740
a awesome amazing helicopter that's like

1117
00:38:50,160 --> 00:38:54,480
holy crap it's flying

1118
00:38:52,740 --> 00:38:56,220
but it is making head roads and that's

1119
00:38:54,480 --> 00:38:58,520
why I'm optimistic that we can change

1120
00:38:56,220 --> 00:38:58,520
this

1121
00:38:59,940 --> 00:39:05,359
cool anybody else

1122
00:39:02,160 --> 00:39:05,359
okay this will be the last question

1123
00:39:07,560 --> 00:39:12,119
in this space What competition does

1124
00:39:10,320 --> 00:39:15,540
Linux have to deal with

1125
00:39:12,119 --> 00:39:19,079
um remembering from early 2000 or early

1126
00:39:15,540 --> 00:39:20,820
to late noughties Windows CE was on a

1127
00:39:19,079 --> 00:39:23,220
bunch of cars and that was just

1128
00:39:20,820 --> 00:39:25,079
horrendous for most part

1129
00:39:23,220 --> 00:39:27,480
I don't know if it's as much a matter of

1130
00:39:25,079 --> 00:39:28,800
competition as it is a matter of

1131
00:39:27,480 --> 00:39:31,560
establishment

1132
00:39:28,800 --> 00:39:32,520
the the biggest thing that you see when

1133
00:39:31,560 --> 00:39:34,200
you look into the decision-making

1134
00:39:32,520 --> 00:39:36,359
process is that they go with what they

1135
00:39:34,200 --> 00:39:38,280
can trust and they trust what they know

1136
00:39:36,359 --> 00:39:40,020
and right now they don't know Linux well

1137
00:39:38,280 --> 00:39:42,599
enough and that's part of the things

1138
00:39:40,020 --> 00:39:44,339
there's a good interview with one of the

1139
00:39:42,599 --> 00:39:46,980
lead development people behind the

1140
00:39:44,339 --> 00:39:50,220
Ingenuity helicopter uh yeah the injury

1141
00:39:46,980 --> 00:39:52,440
helicopter and he actually is a Linux

1142
00:39:50,220 --> 00:39:55,380
person from JPL and he helped get Linux

1143
00:39:52,440 --> 00:39:57,359
into JPL uh it's it's an episode on the

1144
00:39:55,380 --> 00:39:58,980
Linux unplugged podcast it's a quite a

1145
00:39:57,359 --> 00:40:01,140
long interview if you want to go find it

1146
00:39:58,980 --> 00:40:04,020
in their archives and and he outlines

1147
00:40:01,140 --> 00:40:06,240
quite in detail how Linux got into this

1148
00:40:04,020 --> 00:40:08,460
environment and and why he was able to

1149
00:40:06,240 --> 00:40:11,460
get them to trust it and it was a long

1150
00:40:08,460 --> 00:40:13,680
process Linux was in use at JPL for over

1151
00:40:11,460 --> 00:40:16,140
a decade before this project they were

1152
00:40:13,680 --> 00:40:17,579
using Linux on their desktops but for

1153
00:40:16,140 --> 00:40:19,079
the flight Hardware they were compiling

1154
00:40:17,579 --> 00:40:22,740
to these other environments they were

1155
00:40:19,079 --> 00:40:24,359
compiling to things like rtos or any one

1156
00:40:22,740 --> 00:40:25,560
of these other environments like VX

1157
00:40:24,359 --> 00:40:27,300
works is kind of falling out of action

1158
00:40:25,560 --> 00:40:28,619
but it's still quite a dominant one like

1159
00:40:27,300 --> 00:40:31,800
all of the Mars rovers that are still

1160
00:40:28,619 --> 00:40:34,260
moving uh they all run VX works and and

1161
00:40:31,800 --> 00:40:35,700
I don't think that it's wrong to assume

1162
00:40:34,260 --> 00:40:37,920
that critical missions in the future

1163
00:40:35,700 --> 00:40:39,900
will probably also still run VX works

1164
00:40:37,920 --> 00:40:42,000
for a while because VX works is one of

1165
00:40:39,900 --> 00:40:44,940
those supported platforms for the CFS

1166
00:40:42,000 --> 00:40:47,099
they have a history with this they know

1167
00:40:44,940 --> 00:40:49,020
that they can make it do the job and

1168
00:40:47,099 --> 00:40:51,540
until we get to the point where they can

1169
00:40:49,020 --> 00:40:54,240
also feel that confident in Linux

1170
00:40:51,540 --> 00:40:55,980
it won't be in the decision tree and

1171
00:40:54,240 --> 00:40:58,500
that is really where it comes down to

1172
00:40:55,980 --> 00:41:00,839
that organization that testing

1173
00:40:58,500 --> 00:41:02,940
that putting it out on the table and

1174
00:41:00,839 --> 00:41:04,500
saying look we all trust this look how

1175
00:41:02,940 --> 00:41:06,420
much work we're doing to make it so that

1176
00:41:04,500 --> 00:41:08,280
you could too

1177
00:41:06,420 --> 00:41:11,040
and then it will become a matter of well

1178
00:41:08,280 --> 00:41:12,780
why don't we pick Linux because it is

1179
00:41:11,040 --> 00:41:14,640
easier to work with they're already

1180
00:41:12,780 --> 00:41:17,940
using it in so many other places but

1181
00:41:14,640 --> 00:41:19,500
choosing to Crush down into these more

1182
00:41:17,940 --> 00:41:21,119
tightly constrained environments like

1183
00:41:19,500 --> 00:41:22,740
real-time operating systems that don't

1184
00:41:21,119 --> 00:41:25,740
have full posix layers and things like

1185
00:41:22,740 --> 00:41:27,300
that because they trust them more

1186
00:41:25,740 --> 00:41:29,940
and if they don't have to do that work

1187
00:41:27,300 --> 00:41:32,520
well why would they want to

1188
00:41:29,940 --> 00:41:35,099
so I think that it's uh it's a shifting

1189
00:41:32,520 --> 00:41:36,839
trade that will definitely improve over

1190
00:41:35,099 --> 00:41:38,960
time it's not as much a direct

1191
00:41:36,839 --> 00:41:41,099
competition kind of a thing

1192
00:41:38,960 --> 00:41:43,500
case in point they ripped all of the

1193
00:41:41,099 --> 00:41:46,440
windows off the Control software of the

1194
00:41:43,500 --> 00:41:47,700
ISS in about 20 I want to say 12. it was

1195
00:41:46,440 --> 00:41:49,920
in a slide from last year if you're

1196
00:41:47,700 --> 00:41:51,180
curious because someone accidentally

1197
00:41:49,920 --> 00:41:53,820
brought a worm onto the International

1198
00:41:51,180 --> 00:41:56,040
Space Station on a floppy

1199
00:41:53,820 --> 00:41:57,420
that was a bit of a bit of a red you

1200
00:41:56,040 --> 00:41:58,859
know a bit of a black eye and they were

1201
00:41:57,420 --> 00:41:59,940
like you know what

1202
00:41:58,859 --> 00:42:01,320
we're just going to make all of these

1203
00:41:59,940 --> 00:42:03,119
laptops that actually do the system

1204
00:42:01,320 --> 00:42:04,560
displays about what's actually going on

1205
00:42:03,119 --> 00:42:05,640
on the space station Hardware you know

1206
00:42:04,560 --> 00:42:06,900
what they'll just be Debbie and Linux

1207
00:42:05,640 --> 00:42:08,640
now

1208
00:42:06,900 --> 00:42:10,619
because that's good for the job and it's

1209
00:42:08,640 --> 00:42:12,960
on a regular laptop and IBM can provide

1210
00:42:10,619 --> 00:42:15,839
support or someone else it's

1211
00:42:12,960 --> 00:42:17,880
trusted in that niche

1212
00:42:15,839 --> 00:42:19,740
it was good enough even back then

1213
00:42:17,880 --> 00:42:22,520
but it's all about those niches and

1214
00:42:19,740 --> 00:42:22,520
where they can trust them

1215
00:42:23,119 --> 00:42:28,280
what everybody final clip for Sam Bishop

1216
00:42:28,690 --> 00:42:32,449
[Applause]