Weird so many people went for FSMs. You would have thought that a simple literature search would have brought up all sorts useful nuggets. Given the scenario yo

Weird so many people went for FSMs. You would have thought that a simple literature search would have brought up all sorts useful nuggets. Given the scenario you describe - no goal other than avoid collisions, the scheme you implemented is basically a simplified potential field approach. All that being said, I can appreciate the core of this story. I spent my final year implementing, testing, and optimizing D-Lite for a robot (basically a fancy incremental version of A), only to find out that in the end, given our sensor performance, the stock A* planner that came with it did almost nearly as well. Oops.