Experimental Results


Throughout the project, I've kept tabs on my output. Though I'd love to say that the results got better, they were still fairly variable. I think a main reason for that is the map I chose to use [simple.world]. This proved difficult because there are some tight spots that the robots don't deal well with. Especially three when they're trying to follow one another through a corridor. They can get in each others' ways very easily. I changed a few variables to change the follow distance, etc. But I only got so far.

The original map. Starting position was SouthWest corner.
Output with two robots and random walk
Output with two robots and random walk, catch-up code included but jitter occured due to loss of precision
Output with two robots and directed walk, smart metric for termination & jitter reduced
Output with three robots; crowding occurred and crashing/deadlock ensued.

All material & intellectual property on this site
was gathered by Scott Jason Friedman for CS553
at Washington University in St. Louis, 2004.