RRT-based Motion Planning for In-pipe Walking Robots

Abstract

Planning paths on the inner surface of a pipe is one of the many problems that need to be solved in order to make practical use of in-pipe robots possible. In this paper, a motion planning algorithm based on rapidly exploring random trees (RRT) is proposed. The algorithm uses representation of the pipeline as a point cloud, which is easier to generate than the geometric representations required by some of the other existing motion planning algorithms for walking in-pipe robots. The properties of the algorithm and their dependence on the geometry of the pipe are studied.