For the last few years I’ve taken part in the local Rampaging Chariots competition along with a group of graduates. One of our most recent challenges has been to get a rover to autonomously navigate the assigned assault course. So lately I’ve been thinking a lot about odometry, that is, estimating the position of a robot over time. Typically, a cheap method of tracking the position of a rover is by attaching rotary encoders to the output shaft of the drive wheels. Unfortunately as the traction of the drive wheels isn’t perfect, they will slip occasionally and introduce errors into the estimated position.
Computer mice perform a very similar task. By taking low resolution photos a surface they’re able to very accurately track their relative position. If you were able to mount one (or more) of these under a rover, it might prove to be a better odometry solution. (more…)