High-Speed Robot Path Planning in Time-Varying Environment Employing a Diffusion Equation Strategy

Abstract

This paper discusses a method for combined map and sensor-based planning of smooth collision-free robot motion paths. Application of an unsteady diffusion equation model to path planning in a time-varying environment distinguishes this approach from prior work in the field. Collision-free robot paths between a starting and a goal point are generated very rapidly by on-line simulation of the diffusion process and by repetitively evaluating the gradient of the computed concentration distribution functions. Related numerical algorithms can be easily implemented on a massively parallel distributed processor architecture. Applications of this approach to high-speed dynamic path planning for a robot vehicle operating in various types of time-varying 2-D environment are presented. Furthermore, the usefulness of the approach for robot arm path planning in configuration space is demonstrated.