Mission-Level Path Planning for Rover Exploration

Abstract

We present TEMPEST, an autonomous, energy-cognizant planner for long-range rover navigation. Its purpose is to plan sequences of actions that avoid large-scale obstacles, balance energy expenses against available resources, and satisfy operational constraints while attaining a distant goal. An algorithm called Incremental Search Engine (ISE) enables TEMPEST to search over high-dimensional spaces, and provides highly efficient re-planning in response to changes in state transition costs. Models of the world and rover adapt ISE to the rover navigation domain. We propose the concept of Mission-Level Path Planning, then describe how TEMPEST uses ISE to solve for mission-level plans in a spatial, temporal and energy space. A simple application problem aims to illustrate TEMPEST behavior under various time and rover locomotion power conditions.