Comparison of geometrical, kinematic, and dynamic performance of several potential field methods

The artificial potential field (APF) approach is a widely used path-planning algorithm for robot navigation. The method in some of its variants is known to exhibit several difficulties such as entrapment in a local minimum. In this paper, we compare the behavior of several variants of APF methods with emphasis on the quality of path geometry, and velocity and acceleration profiles. We argue that great improvement results when globally consistent potential fields (such as distance to the target) are used in conjunction with boundary following algorithm.