Most autonomous ground vehicle systems that exist today are designed around a single rover. While this is simpler from a coordination perspective, singular robots have issues with highly accurate localization of themselves or their terrain features, and usually require significant human interven- tion in the event they encounter difficulties navigating terrain. These factors combined make using AGVs (henceforth called rovers) expensive and difficult to use for exploration or surveillance of terrain, and the potential risk of losing a (likely very costly) rover in a sand trap or mud pit often precludes the investigation of more interesting topographies.

The goal of this project is to develop a system for rovers to assist other rovers (and human teleoperators) in diagnosis and resolution of terrain traversal problems. Specifically, the project will give rovers the capacity to get themselves out of common terrain traps or vehicle failures like sand traps or high centering, which should enable more robust and adventurous exploration of terrain in many types of planetary environments. As the world moves towards a future with much broader scale rover deployment, direct human control of all rovers will become less and less feasible. This system will solve that problem.

In addition, rover cameras have limited ability to image the rover carrying them, and limited ability to localize features in their environment. With two rovers looking at each other,this system could both improve diagnosis of many kinds of mechanical or system failure on a companion rover that a single rover couldn’t, and provide an immediate method to cross reference localization data.