We show that parallel search techniques derived from their sequential counterparts can enable the solution of motion planning problems that are computationally impractical on sequential machines. We present a parallel version of a robot motion planning algorithm based on “quasi best first” search with randomized escape from local minima and random backtracking, and discuss its performance on a variety of problems and architectures.
|Original language||English (US)|
|Title of host publication||Machine Intelligence and Pattern Recognition|
|Number of pages||13|
|State||Published - Jan 1 1994|
|Name||Machine Intelligence and Pattern Recognition|