WEBVTT

00:00.120 --> 00:03.200
- Velan Engineering Center, and it will be on sensors

00:03.340 --> 00:05.893
for autonomous operations and survivability.

00:10.261 --> 00:12.883
- Thank you, good morning everybody.

00:15.110 --> 00:16.603
The title of my presentation,

00:16.750 --> 00:19.403
Sensors for Autonomous Operations and Survivability.

00:20.650 --> 00:25.630
Often at Night Vision when we talk about autonomous systems

00:26.010 --> 00:30.150
we're referring to RISTA types of sensors

00:31.450 --> 00:32.520
reconnaissance, surveillance,

00:32.520 --> 00:34.070
and target acquisition systems,

00:34.630 --> 00:36.080
and how can we automate them.

00:36.541 --> 00:39.270
But here I'm also suggesting

00:39.270 --> 00:44.270
a more holistic approach to autonomous sensing.

00:44.832 --> 00:49.260
There's in, in robotics there's often a lot of discussion

00:49.260 --> 00:51.193
about autonomous mobility.

00:53.600 --> 00:55.430
But I think all three areas

00:55.910 --> 01:00.433
of not just mobility or maneuver, but maneuver, RISTA,

01:00.585 --> 01:03.750
and engagement would all benefit

01:03.850 --> 01:07.923
from improved automation and situational awareness.

01:14.760 --> 01:19.460
So, this chart shows just a couple of options

01:20.200 --> 01:25.200
on how we might configure a manned, unmanned type of force.

01:29.964 --> 01:34.463
These aren't meant to be, you know the answers,

01:34.530 --> 01:36.080
they're just meant to highlight

01:36.850 --> 01:41.653
some of the pluses and minuses in risk areas.

01:43.600 --> 01:45.100
But you can see in the manned,

01:45.103 --> 01:47.773
or optionally manned platform,

01:48.093 --> 01:51.110
you know we could choose to have a long range sensor

01:51.160 --> 01:54.630
on that vehicle, or as it's shown in the bottom,

01:54.630 --> 01:58.623
perhaps not if we're trying to save money,

01:59.810 --> 02:01.683
perhaps reduce crew workload.

02:03.060 --> 02:04.313
That might be an option.

02:06.120 --> 02:08.880
But if you'll notice in all of the vehicles

02:09.370 --> 02:12.583
we talk about 360 degree situational awareness,

02:13.190 --> 02:14.350
and you know that includes

02:14.350 --> 02:16.513
the long range target acquisition,

02:16.930 --> 02:19.800
but it also includes sensors that are on the vehicle

02:19.800 --> 02:22.740
that would allow the vehicle to understand

02:22.807 --> 02:27.740
what's in it's environment, for example, short range,

02:27.740 --> 02:32.723
say threat forces, or even perhaps friendly forces.

02:32.970 --> 02:34.930
You'd want these vehicles to understand

02:34.930 --> 02:38.429
that there are personnel in either case

02:38.429 --> 02:40.133
in close to the vehicle.

02:41.680 --> 02:46.680
Also, the vehicles need to know if they're close

02:46.680 --> 02:49.453
to other unmanned systems or manned vehicles,

02:50.040 --> 02:51.690
so being able to have

02:51.690 --> 02:55.523
that 360 degree situational awareness is important.

02:56.230 --> 03:01.230
And lastly COMs, the COMs bandwidth

03:02.130 --> 03:03.883
and the availability of that,

03:03.940 --> 03:08.940
is it going to be a large driver in many of these options,

03:11.160 --> 03:16.160
for example if you don't have an automated,

03:16.357 --> 03:21.357
say search and detection system on your robotic platform,

03:22.706 --> 03:25.380
that's going to require a high-bandwidth link

03:25.400 --> 03:27.580
back to the manned platform for an operator

03:27.580 --> 03:30.293
in that vehicle to be able to search for a threat.

03:30.840 --> 03:33.460
We'd like to automate that as much as possible

03:33.577 --> 03:36.063
to reduce that bandwidth requirement.

03:43.000 --> 03:48.000
You know today the humans are responsible for conducting

03:48.890 --> 03:52.997
all of the functions in the platform, maneuver,

03:54.655 --> 03:59.655
command and control, surveillance, engagement.

04:01.590 --> 04:04.540
What we can do in the near term, or what we'd like to do

04:04.990 --> 04:07.950
is have that same vehicle control

04:07.950 --> 04:10.723
one or more robotic systems.

04:11.003 --> 04:12.683
And we can do that,

04:13.521 --> 04:16.531
you know we have drive by wire capabilities now,

04:16.531 --> 04:20.173
an auto loader for engagement,

04:20.920 --> 04:22.670
we can do supervised teleoperation,

04:23.210 --> 04:25.683
or semi-autonomous navigation,

04:26.431 --> 04:29.653
but we need automation for all those functions.

04:30.490 --> 04:32.590
In the future, you know we'd like to have

04:32.590 --> 04:34.913
a fully autonomous system that's intelligent.

04:35.370 --> 04:37.143
That can do a lot of this work,

04:37.150 --> 04:41.203
it doesn't require a constant interaction by an operator.

04:45.300 --> 04:50.300
To really gain the advantage in the future engagement,

04:52.700 --> 04:55.400
we need to provide this automation

04:55.400 --> 04:59.653
that can help augment our force to think and act faster.

05:00.470 --> 05:03.373
Military environments are extremely complex.

05:05.603 --> 05:07.210
You know we need solutions

05:07.210 --> 05:09.423
beyond just what is commercially available.

05:10.030 --> 05:12.823
I mean there's a lot of commercial technology out,

05:14.010 --> 05:18.563
available these days, algorithms for detection,

05:18.683 --> 05:21.603
facial recognition, things like that.

05:22.180 --> 05:25.040
But you have to remember in a military context

05:25.430 --> 05:28.083
we are in an adversarial environment.

05:28.570 --> 05:31.793
People don't want to be seen, they don't,

05:33.911 --> 05:35.911
you don't want a vehicle to be detected.

05:37.390 --> 05:40.350
So the better of we can basically be able to see

05:40.350 --> 05:43.253
and understand what's going on,

05:43.370 --> 05:45.650
in that type of adversarial environment

05:47.184 --> 05:48.693
would be a big benefit.

05:51.330 --> 05:56.330
Alright, so on this chart I kind of highlighted

05:57.880 --> 06:02.403
that there's the lethality, the mobility,

06:02.900 --> 06:04.520
and the surveillance component

06:06.800 --> 06:09.050
are typically kind of how we break down

06:09.520 --> 06:12.203
the three main capabilities on a platform.

06:12.810 --> 06:17.810
And that how we're doing a lot of that today in the field,

06:18.230 --> 06:20.563
it really comes down to the manned operator,

06:22.606 --> 06:26.523
but we'd like to make that more autonomous.

06:27.690 --> 06:32.520
And you know the same analogy I guess,

06:32.520 --> 06:36.823
with commercial technology could apply here as well.

06:37.010 --> 06:37.993
You know we have,

06:38.300 --> 06:41.063
you've heard about autonomous driving vehicles,

06:41.552 --> 06:44.743
you know whether it's an Uber, or a Tesla,

06:46.450 --> 06:47.830
you know but they're really,

06:47.830 --> 06:52.053
out of the six levels of autonomy,

06:52.270 --> 06:55.003
they're at about a half-way point right now.

06:55.730 --> 06:58.940
So even driving on a road in a structured environment

06:59.284 --> 07:01.600
with a myriad of sensors

07:01.600 --> 07:04.263
on those platforms it's still a challenge.

07:05.101 --> 07:09.573
The same exists in the surveillance world.

07:11.190 --> 07:13.610
You know the automation of these functions

07:13.610 --> 07:18.523
of search, target detect, tracking, classification, ID,

07:18.550 --> 07:22.500
hostile fire detection really have never been done

07:22.500 --> 07:24.520
at this kind of level before,

07:24.520 --> 07:26.453
combining all those capabilities.

07:26.530 --> 07:27.890
Much work has been done

07:28.310 --> 07:33.310
on individual types of algorithms but not combined.

07:35.640 --> 07:38.603
You know we used to say many years ago that,

07:39.000 --> 07:40.630
you know we needed 300 pixels

07:40.630 --> 07:43.993
on a target to do an ID function.

07:45.644 --> 07:47.170
I'm not sure if we said that

07:47.170 --> 07:48.770
because that's what we could do,

07:48.910 --> 07:50.470
or that's what we really wanted,

07:50.470 --> 07:53.943
but I think we figured out that we need more like 2000.

07:54.940 --> 07:59.940
So work going on with focal plane arrays can provide that.

08:04.000 --> 08:07.620
And autonomy and automation,

08:07.620 --> 08:09.683
you know it's not just for robots.

08:10.254 --> 08:14.460
We can use these algorithms inside of the vehicle

08:14.560 --> 08:17.970
to reduce the operator workload, to allow him to do more

08:19.890 --> 08:21.713
other than just his one job.

08:26.370 --> 08:28.780
So this is something that we're working on.

08:28.780 --> 08:32.720
We realize these challenges, at CERDEC Night Vision

08:32.850 --> 08:36.350
and we're aggressively investing in technology

08:36.350 --> 08:40.033
to try to help solve some of these problems.

08:41.590 --> 08:44.940
And one of the ways to do this is to take advantage

08:44.940 --> 08:47.190
of all the information that's on the vehicle.

08:48.140 --> 08:50.940
Typically in the past, you know an ATR, it was just,

08:50.940 --> 08:54.483
say it was a basically before a RISTA system,

08:54.790 --> 08:59.790
it just used that high-performance cooled IR camera

09:00.210 --> 09:03.203
and ran detection algorithms.

09:04.750 --> 09:06.923
We'd like to include other information.

09:07.340 --> 09:08.803
What about DTED data?

09:09.510 --> 09:12.040
What about other information,

09:12.040 --> 09:15.980
for example command and control

09:15.980 --> 09:17.543
threat detection information.

09:19.050 --> 09:22.230
A good example would be if you understand the terrain

09:23.150 --> 09:25.453
you may have a detection from an algorithm,

09:25.460 --> 09:30.337
might be a false alarm that it looks like a tank is say,

09:30.980 --> 09:33.403
in an area, say a tank is in a tree.

09:33.720 --> 09:36.760
Where we wouldn't expect that to actually happen.

09:36.760 --> 09:38.570
So you could use that information.

09:38.570 --> 09:41.203
Not just rely solely on the algorithm itself.

09:41.947 --> 09:46.947
But then we'd like to look more at understanding the scene

09:49.840 --> 09:52.743
and what is going on in the scene itself.

09:53.110 --> 09:56.163
So more than just the algorithm,

09:56.240 --> 10:00.073
trying to gain some understanding, and then reasoning,

10:01.380 --> 10:04.510
and then providing threat detection information

10:04.510 --> 10:07.113
that's more in context.

10:07.200 --> 10:09.070
Giving the operator information

10:09.070 --> 10:12.573
about what exactly the algorithm thinks it's seeing,

10:13.000 --> 10:16.433
not just a detection all by itself.

10:20.520 --> 10:25.520
So lastly, I mentioned this in the introduction,

10:26.152 --> 10:31.152
I believe that these autonomous capabilities

10:32.160 --> 10:35.143
support all three areas of mobility,

10:35.330 --> 10:36.733
lethality, and surveillance.

10:37.232 --> 10:41.773
If you had DTED maps on a vehicle for autonomous mobility,

10:42.045 --> 10:45.450
those same DTED maps as I just mentioned could help

10:45.450 --> 10:48.573
with surveillance and ATR algorithms,

10:48.973 --> 10:50.513
could also help in lethality.

10:51.530 --> 10:55.567
So the idea of having a single type

10:56.700 --> 11:00.150
of situational awareness capability

11:00.150 --> 11:03.163
that supports all three of these areas.

11:03.760 --> 11:06.860
For example, autonomous mobility system,

11:06.860 --> 11:08.210
I'm sure would like to be aware

11:08.210 --> 11:11.543
that there's a threat vehicle located,

11:11.740 --> 11:14.033
that was located by the surveillance system.

11:14.240 --> 11:17.380
We would want to maneuver out of contact

11:17.380 --> 11:18.733
with that threat platform.

11:19.060 --> 11:21.780
So the ability to have all of these things working together,

11:21.780 --> 11:22.833
I think is key.

11:24.230 --> 11:28.477
We also think that embedded processing at the sensor

11:28.477 --> 11:32.653
is also another key factor.

11:33.130 --> 11:35.950
I mean we all look at our cell phones

11:35.950 --> 11:38.663
and we understand how small the processing can be.

11:40.170 --> 11:42.197
If we are going to have more pixels,

11:42.197 --> 11:47.197
we can't be transmitting all of this data

11:47.530 --> 11:50.770
across the vehicle, some of that has to be processed

11:50.770 --> 11:52.170
at the sensor itself

11:54.511 --> 11:56.533
to be more practical.

11:58.350 --> 12:01.550
And then we talked a lot about intelligent sensors

12:01.945 --> 12:06.945
but we'd like to combine the functions of sensors into one.

12:07.490 --> 12:10.423
So if you have a driving camera,

12:10.600 --> 12:12.100
we'd also like that same camera

12:12.100 --> 12:14.143
to be used for hostile fire detection.

12:14.531 --> 12:16.380
With the same cameras can be used

12:16.380 --> 12:19.701
all around the vehicle and can run algorithms

12:19.701 --> 12:23.113
to provide that autonomous capability.

12:26.680 --> 12:29.450
And lastly we are interested

12:29.450 --> 12:31.303
in an interesting engagement here.

12:33.470 --> 12:36.613
You know there is a lot of SNT that still needs to be done.

12:37.240 --> 12:41.583
There's a lot of components that are in play.

12:42.160 --> 12:44.360
There's a lot of work that's being done now.

12:44.610 --> 12:48.193
But we're always interested in more ideas,

12:48.193 --> 12:52.210
innovation of course, and your know-how.

12:52.210 --> 12:57.210
So that concludes my talk, pending any questions.

13:03.610 --> 13:04.623
Okay, sir.

13:09.670 --> 13:13.500
- [Man] I was wondering if you got fully autonomous vehicles

13:13.500 --> 13:16.330
that are basically engaging the enemy

13:16.330 --> 13:18.523
or in hostile territory.

13:19.293 --> 13:22.550
There are gonna be decisions that that platform

13:22.550 --> 13:26.450
has to make that in the human realm of endeavor

13:26.450 --> 13:30.233
would be considered perhaps ethical or moral.

13:30.700 --> 13:34.710
So do you see a way forward to, I guess,

13:34.710 --> 13:36.430
enview that autonomous system

13:36.430 --> 13:41.387
with any of those more human attributes or characteristics?

13:42.870 --> 13:46.773
- Yeah, I think for the foreseeable future,

13:47.340 --> 13:49.470
I think those decisions are going to require

13:49.470 --> 13:52.440
a man in the loop, always.

13:52.440 --> 13:56.403
I mean, perhaps long-term,

13:57.599 --> 14:00.940
an autonomous system might be able to understand

14:01.360 --> 14:05.383
rules of engagement, but I think,

14:05.400 --> 14:09.680
you know until it's proven, it's just gonna take some time.

14:11.640 --> 14:12.940
- [Man] Human in the loop.

14:13.070 --> 14:14.353
- Yes, human in the loop.

14:14.890 --> 14:16.140
- [Man] That makes sense.

14:16.380 --> 14:18.210
I just keep thinking of the movie I, Robot,

14:18.210 --> 14:20.103
and it's not a pretty thought.

14:20.470 --> 14:21.303
- Right, right.

14:27.800 --> 14:29.460
- [Woman] I understand that you're wanting

14:29.460 --> 14:31.580
to get these technologies out quicker

14:31.580 --> 14:34.420
and that possibly our enemies are getting them quicker

14:34.420 --> 14:38.163
using commercially available parts.

14:38.390 --> 14:41.920
So from a component manufacturer perspective

14:42.159 --> 14:46.360
is the Army going to be looking for more commercial units

14:46.360 --> 14:49.250
or still holding to the traditional mill standards

14:49.250 --> 14:52.003
and specific industry regulations?

14:52.080 --> 14:53.630
Is there gonna be any relaxing

14:53.630 --> 14:55.223
on those sort of requirements?

14:59.490 --> 15:03.171
- I mean I think we're willing to do,

15:03.171 --> 15:05.333
in order to get things out quicker,

15:05.480 --> 15:06.910
I think we're willing to look

15:06.910 --> 15:10.163
at alternative approaches for that.

15:10.430 --> 15:11.970
And if that requires,

15:11.970 --> 15:15.370
you know reducing some of the requirements,

15:15.370 --> 15:17.700
the military specific requirements.

15:17.700 --> 15:21.340
There will always be some things that we have to adhere to

15:21.562 --> 15:25.970
but I think yes we are interested

15:25.970 --> 15:27.793
in getting things out quicker.

15:28.120 --> 15:31.203
So we'd be looking for how to better do that.

15:34.130 --> 15:38.350
Okay, alright, thank you.

15:41.652 --> 15:44.569
(group applauding)