Empirical Sampling of Path Sets for Local Area Motion Planning - Robotics Institute Carnegie Mellon University

Empirical Sampling of Path Sets for Local Area Motion Planning

Conference Paper, Proceedings of International Symposium on Experimental Robotics (ISER '08), pp. 451 - 462, July, 2008

Abstract

We consider the problem of online planning for a mobile robot among obstacles, where it is impractical to test all possible future paths. One approach is for the runtime task to test some subset of the possible paths and select a path that does not collide with obstacles while advancing the robot toward its goal. Performance depends on the choice of path set. In this paper we assume the path set is fixed and chosen offline. By randomly sampling the space of path sets we discover effective path sets?omparable or superior to the best previously suggested approaches. In addition, testing large numbers of randomly generated path sets yields some insights on the relation of robot performance to the choice of path set.

BibTeX

@conference{Knepper-2008-10037,
author = {Ross Alan Knepper and Matthew T. Mason},
title = {Empirical Sampling of Path Sets for Local Area Motion Planning},
booktitle = {Proceedings of International Symposium on Experimental Robotics (ISER '08)},
year = {2008},
month = {July},
pages = {451 - 462},
publisher = {IFRR},
keywords = {motion planning, nonholonomic, path, trajectory, sampling, diversity, separation},
}