A near-optimal sensor-based motion-planning algorithm for parts mating

Abstract

We propose a near-optimal sensor-based motion-planning algorithm for parts mating. In part mating, only a deadlock-free path to a target point exists around the configuration space obstacle of a point robot. Because the number of unknown obstacles is only one, our algorithm is able to seek for a complicated deadlock-free path efficiently, whose upper bound of the worst path length is exactly evaluated by p+2D (p: the perimeter of the configuration space obstacle; and D: the Euclidean, distance between initial and target points). In this type of sensor-based motion-planning, the lower bound of the worst path length is described by p+D. On the other hand, an upper bound of the worst path length was given by 1.5p+D in the previous algorithm Bug1, and also it was obtained by 2p+D in the previous algorithms Alg1 and Alg2. Comparing our new algorithm with these previous algorithms, if D is fully smaller than p, we ascertain our algorithm's superiority theoretically and experimentally. The proposed algorithm is a significant algorithm because there is only a configuration space obstacle for practical parts mating.