Shared publicly  - 
Mars as seen through MAHLI (MSL's microscope), and MAHLI as MSL sees it.
Scott Maxwell's profile photoAndy Dillon's profile photo
So, how is straight line navigation (or similar) maintained over unmapped terrain, when using a programmed drive sequence?
+Andy Dillon Sometimes we don't care because the terrain is benign enough that as long as the rover's going in the right direction, we're happy.

Often, of course, we do care. For those times, our usual trick is visual odometry. The rover takes a 3-D picture, moves a little (not usually more than a meter), and then takes another 3-D picture. The rover finds common 3-D points in the images, and does some math to figure out from their apparent motion how it (the rover) must actually have moved. It corrects its notion of its position, and the cycle repeats.

So that tracks the rover's position. The rover's heading is furnished by an inertial measurement unit (IMU), which is like a compass that doesn't use magnets (because Mars's magnetic field is too weak).

With the combination of these two elements -- the IMU to keep your heading straight, and visual odometry to correct your position -- you can progress in a straight line (or along any other path) as far as you want!
Sweet - thanks so much for the explanation, Scott! I had wondered if there was an iterative landmark-based visual component, but had not imagined the inertial component.
Add a comment...