Improved hierarchical planner performance using local path equivalence

Abstract

We propose a motion planning algorithm that reasons about tradeoffs between the discrete decision problems and continuous optimization problems faced by a mobile robot navigating through a cluttered environment. Discrete decisions typically involve transient options as the robot selects corridors to traverse. By contrast, optimization occurs within open spaces among homotopic paths. We utilize properties of local path sets to detect decisions of immediate importance and select routes that maximize the chance of future success.