Shared publicly  - 
 
Today on MSL, we sequenced the arm even though we didn't have data from the approach drive to the target on the lip of Yellowknife Bay.

We were supposed to have gotten the data from the drive, of course; we wouldn't normally sequence the arm without it. But there was some kind of problem -- I never nailed down the specifics -- where the rover thought the post-drive imaging was transmitted, but it got stuck on the relay orbiter, or something. Anyway, we didn't have it on the ground.

What we did have was the previous sol's imagery from before the drive, and coordinates from the end of the drive. These coordinates were about as good as gold, because they were corrected by visual odometry.

(Visual odometry? That's where the rover takes a stereo pair of images, moves a little, and then takes the images again. By finding how common 3-D points moved between the first and second pair of images, the rover can mathematically work out how it must have actually moved. If it slipped a little, say -- instead of going 1 meter, it got a little bit bogged down in sand and went only 90 cm -- it'll correct its notion of its position. You do this unconsciously all the time when you walk around; it's why you don't bump into things. Well, not much.)

So we took the new coordinates, plugged them into the previous day's images, and got sequencing.

Now, we'd never actually send such a sequence to the rover. We need that post-drive imaging to confirm exactly where things are in the rover's world before we uplink any commands to its arm. We worry about matters of centimeters, sometimes even millimeters; there's no substitute for the absolute precision of the post-drive imaging. But this approach was good enough to get the vast bulk of the work out of the way, keeping us more or less on schedule.

We did get those images later in the day; as we'd hoped it would, the rover re-sent them during its morning communication pass, so we got them in our afternoon. A few small corrections later, and the sequence was essentially ready to go. In the end, we ran maybe an hour late on the planning day -- but if we hadn't taken the chance, we wouldn't have gotten anything done at all; we'd have had to just punt.

We dare mighty things. And sometimes, we dare smaller things, and here and there it saves us a sol. That's how it goes in the space biz.
14
1
David Ford's profile photo
 
nice work and nice explanation
Add a comment...