From cross-country autonomous navigation to intelligent deep space communications: visual sensor processing at JPL

Abstract

We describe ongoing work at JPL in two fields: autonomous navigation for terrestrial vehicles, and prioritized progressive transmission for deep space communications. While such applications may seem rather disparate, they have in common the need for autonomous reasoning about visual information. We first review a number of techniques currently under development that make use of data from color and infrared cameras, multispectral sensors, and laser rangefinder for estimating properties of the terrain cover in outdoor vegetated terrain. We then discuss how onboard visual analysis mechanisms for Mars rovers can be used for prioritizing the data to be transmitted to Earth, in order to maximize the science return of a mission.