作者: Jory Denny , Nancy M. Amatoo
DOI: 10.1007/978-3-642-36279-8_18
关键词:
摘要: Motion planning has received much attention over the past 40 years. More than 15 years have passed since introduction of successful sampling-based approach known as Probabilistic RoadMap Method (PRM). PRM and its many variants demonstrated great success for some high-dimensional problems, but they all level difficulty in presence narrow passages. Recently, an called Toggle been introduced whose performance does not degrade 2-dimensional problems with In PRM, a simultaneous, coordinated mapping both C free obst is performed every connection attempt augments one maps – either validating edge current space or adding configuration ’witnessing’ failure to other space. this paper, we generalize d-dimensions show that benefits continue hold higher dimensions. particular, introduce new passage characterization, α-e-separable passages, which describes types passages can be successfully mapped by PRM. Intuitively, are arbitrarily regions separate , at least locally, such hallways office building. We experimentally compare methods variety scenarios different robots up 16 dof.