1
00:00:06,480 --> 00:00:14,000
Bennu has a lot of challenges for sample 
collection. The one that we focused on the  

2
00:00:14,000 --> 00:00:21,360
most before launch was the actual sample head 
and our kind of vacuum cleaner to get the dust  

3
00:00:22,000 --> 00:00:28,480
into a head when there's no vacuum or no air 
to help you. So, we've got that one we think.  

4
00:00:28,480 --> 00:00:35,120
And then when we arrived and we saw how rocky 
that surface was, that all of a sudden gave us  

5
00:00:35,120 --> 00:00:42,080
a whole slew of other challenges which basically 
fall into the categories of precision  

6
00:00:42,080 --> 00:00:47,360
navigation and maneuver performance. So, we've 
got a landing site we've got a hit that's now  

7
00:00:47,360 --> 00:00:52,480
five times smaller. We've got huge boulders 
the size of three-story buildings parked on  

8
00:00:52,480 --> 00:00:58,640
the edge of that landing site, so getting in and 
out safely has also become quite a challenge.

9
00:01:02,800 --> 00:01:09,520
The navigation on O-REx for descent 
to the surface is done on board.  

10
00:01:10,160 --> 00:01:17,200
It's an optical navigation system. It involves 
kind of three different teams of people here,  

11
00:01:18,160 --> 00:01:23,600
both here, in Arizona, Goddard people are 
involved. It's a huge bunch of people. So,  

12
00:01:23,600 --> 00:01:30,720
the science guys with all the mapping have built 
what amounts to a 3D map of that asteroid down to  

13
00:01:31,360 --> 00:01:36,560
centimeter precisions, and then we 
figure out the flight path to get to  

14
00:01:36,560 --> 00:01:43,040
the surface. And then along that flight 
path we pick out features to recognize,  

15
00:01:43,040 --> 00:01:48,720
to use as guide points to tell us where we are 
on the way down, and that we're on track, and  

16
00:01:48,720 --> 00:01:54,880
we have two opportunities to correct the course; 
that's the checkpoint and the match point burns.  

17
00:01:55,920 --> 00:02:01,760
One corrects us to the tag site, and one stops 
the relative motion and takes us straight down.  

18
00:02:02,320 --> 00:02:06,880
So the optical system figures out where we are 
well enough to do those burns. Those burns are  

19
00:02:06,880 --> 00:02:15,840
computed on the fly in real time. So we put 
pieces of that digital asteroid map on board.  

20
00:02:15,840 --> 00:02:20,880
The software on board is smart enough to account 
for lighting configurations and everything else.  

21
00:02:21,760 --> 00:02:26,000
It looks at where it's supposed to be, 
says "well, my camera's pointing this way,  

22
00:02:26,720 --> 00:02:32,640
that feature ought to be up here, I don't know 
precisely where I am." There's an uncertainty,  

23
00:02:32,640 --> 00:02:35,840
so it draws a box around where the feature 
ought to be and starts looking for it.  

24
00:02:36,640 --> 00:02:40,480
One feature will give you a direction. 
Multiple features give you direction,  

25
00:02:40,480 --> 00:02:45,760
position, the whole nine yards. So we're looking 
for up to five features in each image to tell us  

26
00:02:45,760 --> 00:02:53,280
precisely where we are in three space, and then 
we can also use changes in position to do velocity  

27
00:02:53,280 --> 00:02:56,960
and figure out how we're moving and propagate 
the position. And that's how we get down there.

28
00:03:01,520 --> 00:03:07,120
The whole process of taking the 
images, figuring out where they are,  

29
00:03:07,120 --> 00:03:13,120
correlating them to onboard stuff, that's the 
natural feature tracking system that was developed  

30
00:03:13,840 --> 00:03:18,160
as, actually, a backup to the lidar system 
that was put in place before launch.

31
00:03:22,560 --> 00:03:29,920
So, the surface of Bennu is extremely rough, 
much rougher than we had planned on. It does  

32
00:03:29,920 --> 00:03:35,120
pose hazards to the vehicle in the 
process of getting that sample.  

33
00:03:36,160 --> 00:03:43,600
They fall into two basic categories: there's a 
tip over hazard and there's a back away hazard.  

34
00:03:43,600 --> 00:03:49,120
So, when we come down we're going to touch the 
surface. The head's likely to move because we  

35
00:03:49,120 --> 00:03:53,680
won't be perfectly straight down. There will be 
some lateral velocity. The head's going to stick,  

36
00:03:53,680 --> 00:04:00,480
the vehicle is going to tip, the controller will 
try to hold us still, but it won't be completely  

37
00:04:00,480 --> 00:04:07,040
successful, depending on how fast we actually 
touch, so there is a hazard to the arm that  

38
00:04:07,040 --> 00:04:12,560
if there's a boulder nearby, we could lean over 
into it with the arm. The arm has high pressure  

39
00:04:12,560 --> 00:04:18,080
gas bottles on it. It's got plumbing for the 
gas to feed the vacuum cleaner; it's got pyros  

40
00:04:18,080 --> 00:04:21,840
on it which are basically explosives. You 
don't want to hit that arm on something.  

41
00:04:22,400 --> 00:04:29,920
So that's one hazard. There is also a hazard that, 
having touched, gotten the sample, tipped over,  

42
00:04:29,920 --> 00:04:36,720
now we decide to get out, we're going to back away 
at the attitude we're at at the time, which may be  

43
00:04:36,720 --> 00:04:42,320
a little off from straight up and down, and if we 
tip over too far and then back away, because of  

44
00:04:42,320 --> 00:04:48,480
the size of the obstructions around the TAG site, 
we could fly into something. So there's been a lot  

45
00:04:48,480 --> 00:04:54,960
of analysis of the site using the really high 
precision digital elevation maps and all the  

46
00:04:54,960 --> 00:05:01,200
slopes of the facets and which way will we tip and 
what might we hit depending on which way we tip.  

47
00:05:01,920 --> 00:05:07,688
All that's gone into convincing 
ourselves that we can get down and get out safely.

48
00:05:11,920 --> 00:05:17,440
So, the head comes into contact with the 
surface. There's a spring in the arm so  

49
00:05:17,440 --> 00:05:22,080
that the vehicle will keep pushing towards 
the surface. The spring will soak that up.  

50
00:05:22,080 --> 00:05:28,480
It holds the head in contact with the ground, 
we fire the gas bottles, the bottles will stir  

51
00:05:28,480 --> 00:05:35,280
up the dirt under the head. We only give the 
gas one way out and that's through the vents  

52
00:05:35,280 --> 00:05:42,080
in that sample head, so the stirred up dirt is 
carried with the gas. The gas goes out the vents,  

53
00:05:42,080 --> 00:05:47,200
the dirt is kept by screens in the head, 
and that's how we collect the sample.

54
00:05:51,440 --> 00:05:58,000
We just got finished with Match Point rehearsal, 
and we've used that imagery to estimate,  

55
00:05:58,800 --> 00:06:04,240
based on where we look like we're going 
to come down and what's under us, how much  

56
00:06:04,240 --> 00:06:10,240
sample we might be able to pick up based on the 
surface properties and all the properties of the  

57
00:06:10,240 --> 00:06:17,760
that we see, and the estimate that came in after 
match point rehearsal from the science team was  

58
00:06:18,720 --> 00:06:25,520
around 100 grams, maybe more depending on 
things that we can't tell from imagery like  

59
00:06:25,520 --> 00:06:31,840
how much will the surface compress, what's under 
the particles we can see, but the current guess is  

60
00:06:31,840 --> 00:06:38,400
around 100 grams if we were to come down right 
where we were headed at match point rehearsal.

61
00:06:43,520 --> 00:06:51,680
The sample collection measurement is done 
using a change in momentum of the spacecraft.  

62
00:06:52,720 --> 00:06:57,040
So, if you think about ... the analogy 
I usually use is an ice skater,  

63
00:06:57,920 --> 00:07:03,440
arms out - spins slowly, arms in - spins 
faster. So, that's a momentum issue. We  

64
00:07:03,440 --> 00:07:09,040
have done ... we just do two positions of 
the spacecraft: we do the arm straight out,  

65
00:07:09,040 --> 00:07:14,560
we do the arm straight up. We spin the spacecraft, 
and we see what it takes to spin the spacecraft,  

66
00:07:14,560 --> 00:07:18,800
and we can make an estimate of the inertia 
of that spacecraft with an empty head  

67
00:07:19,600 --> 00:07:22,640
and we know it's completely empty right 
now because we haven't touched a thing.  

68
00:07:23,200 --> 00:07:29,920
So we do that measurement before we go down. After 
we come back up, we do the same two positions,  

69
00:07:30,960 --> 00:07:35,680
and based on the change in that estimated 
inertia, we can tell ... we can estimate  

70
00:07:35,680 --> 00:07:41,046
how much mass must be sitting in the head 
because it's the only thing that's changed.  

71
00:07:44,400 --> 00:07:51,440
We can take three sample attempts from actual 
contact. We have three bottles we can fire, so we  

72
00:07:51,440 --> 00:07:59,360
can basically run the vacuum cleaner three times. 
We can do any number of aborts part way down  

73
00:07:59,360 --> 00:08:03,920
and get out and still have our three attempts. 
Those don't use up a bottle. We have lots of fuel.  

74
00:08:05,040 --> 00:08:10,320
We can add sample into the head. It was 
built for that, so there's no problem  

75
00:08:10,320 --> 00:08:17,760
there if we get a little bit but we want more 
we can go back. There is uncertainty as to  

76
00:08:18,800 --> 00:08:25,600
if we do fire the bottle, the bottle has a huge 
amount of energy in it, given the rocky surface  

77
00:08:25,600 --> 00:08:32,160
and how hard it was to find sampleable spots, we 
don't know what firing that bottle is actually  

78
00:08:32,160 --> 00:08:39,600
going to do to the sampleability of the site. 
So there is a chance that if we fire the bottle  

79
00:08:39,600 --> 00:08:45,280
and we want to go back, the site may or 
may not still be suitable for sampling.

80
00:08:49,760 --> 00:08:55,520
So, even if we have to abort on the first attempt 
because we think we're coming down on a hazard,  

81
00:08:55,520 --> 00:09:02,720
and we have the ability to know that, we have a 
backup site, Osprey, that is in a lot of ways it's  

82
00:09:02,720 --> 00:09:08,960
a little easier to get to. It has fewer hazards 
within the site. It has some on the edge but none  

83
00:09:08,960 --> 00:09:14,720
really within the site, but it doesn't look as 
sampleable. That's why it's the backup, but we do  

84
00:09:14,720 --> 00:09:21,120
have that option, so we aren't going to take ... 
we aren't going to go for broke on the first shot  

85
00:09:21,120 --> 00:09:25,840
and risk the vehicle when we know we have the 
equipment to take more than one sample attempt.

86
00:09:30,160 --> 00:09:37,280
The most exciting part for me has been kind of 
rising to the challenge that Bennu presented, and  

87
00:09:37,920 --> 00:09:41,360
you know, working with all 
the various parts of this team  

88
00:09:41,920 --> 00:09:49,520
to figure out how we're going to change this 
thing, from going to a 25 meter beach to  

89
00:09:49,520 --> 00:09:55,120
a five meter pile of rocks with buildings parked 
around it, and you know, watching how these guys  

90
00:09:55,120 --> 00:10:03,840
have figured out ways around pretty much every 
problem that's come along. It's been a lot of fun.

91
00:10:04,554 --> 00:10:22,420
[Room Tone]