Recursion-Based Probabilistic RoadMap for Robot Path Planning

Abstract:
The probabilistic roadmap (PRM) has become one of numerous robot path planning algorithms that have enjoyed a wide variety of applications. This paper introduces a variant of this method, here dubbed the recursive probabilistic road map (r-PRM). The main difference lays in how the road map is created. Otherwise, the two methods produce essentially equivalent results, so that their choice is mostly a matter of personal preference. Our preference is for r-PRM as it is conceptually simpler and easier to understand. In addition, we have also found that r-PRM typically is faster than PRM. The paper presents full descriptions of both algorithms, so as to explain their differences, and provides the results of various runs of these algorithms in a maze-like environment. It also provides a statistical comparison of their respective performance and behaviors.