Scaling the dynamic approach to autonomous path planning: planning horizon dynamics

Abstract

In the dynamical systems approach to robot path planning both sensed and remembered information contribute to shape a nonlinear vector field that governs the behavior of an autonomous agent. Such systems perform well with partial knowledge of the environment and in dynamically changing environments. Nevertheless, it is a local heuristic approach to path planning, and it is not guaranteed to find existing paths. We describe a method of adjusting the spatial resolution of the planner using a dynamical system that operates at a faster time scale than the planning dynamics. This improves the system's ability to utilize both sensed and remembered information, and to solve a larger range of problems without resorting to global path planning. 1