Heuristic approach to dynamic motion planning for multi-link planar manipulators in a sparse, time varying environment

Abstract

In this paper a fast and simple method to solve the motion planning problem for planar manipulators is suggested. This method is based on the development of a set of simple reactive behaviours, which allow the robot to move through a time-varying workspace. These simple reactive behaviours are activated by each link of the robot in response to the sensory information experienced as the robot moves through its workspace in search of its goal. As a consequence of the robot's behaviours being reactive, the influence of obstacles on the robot's path is only local. However, there is a subset of problems for which sequential movement of links fails to guide the robot to its goal. This constitutes what we describe as a local minima problem. This problem has been solved by producing a behaviour which identifies local minima and defines an intermediate goal for the robot, in order to guide it out of the local minima. Results are given in the paper which illustrate the performance of this planner.