Possible Future WorkMy multi-robot strategy can be extended in a number of ways, but I'll just extrapolate on one. It makes sense to call it the B-tree mapping strategy since that is how the robots are organized structurally. Each Rover is responsible for some number N >= 1 of robots. My current implementation reflects B-tree(1) mapping. Instead of the robots stretching out like a snake around corners and advancing the rear as we proceed, we stretch out like a fan, careful to keep each of the children from traversing the same path. A problem with this approach is that it is not efficient for small, maze-like maps with cramped corridors; it is, however, efficient for large maps with solid obstacles. I would also recommend it for robots that are numerous, cheap, and failure-prone.
was gathered by Scott Jason Friedman for CS553 at Washington University in St. Louis, 2004. |