Efficient local sampling for motion planning of a robotic manipulator

Abstract

Most of the current motion planning algorithms for robotic manipulators are designed to work within the robot's configuration space using an offline global planning method. This restricts their usage to a priori known environments. Using Artificial Potential Fields (APFs) for local planning allows for real-time motion planning in unknown environments and are used successfully for point-mass robots. However, local minima are a well-documented problem associated with APFs. When APFs are applied to more complex robots, such as robotic manipulators, these local minima arise more frequently and they greatly reduce the effectiveness of the APF approach. This paper identifies one such local minima problem specifically related to robotic manipulators, the reach-around local minima, and presents a variant of the subgoal selection method, based on local sampling of the configuration space, to solve it. Simulation results are presented for a number of scenarios demonstrating the efficacy of the proposed approach. © 2012 Springer-Verlag.