The Comparison of Motion Planners for Robotic Arms

摘要

This paper presents a comparison of two motion planning approaches for industrial manipulators. The first approach implements sampling-based methods and the second approach uses optimization-based methods. Specific algorithms compared in this article are RRT, RRTConnect (sampling-based motion planning) and STOMP (optimization-based motion planning). Both approaches are able to find collision-free trajectory and are available for Robotic Operating System via MoveIt package. In our experiments computation time, success rate and trajectory length are evaluated. Our results show that RRT and RRTConnect are faster, but a risk exists, that the length of trajectory will be extremely long. On the other hand, STOMP is slower, but the length of found trajectory depends on parametrization.