Motion planning for all-terrain vehicles: a physical modeling approach for coping with dynamic and contact interaction constraints

Abstract

Addresses modeling and global motion planning issues for an autonomous wheeled mobile robot moving on an uneven three-dimensional terrain. We focus particularly on the issue of dealing with dynamic and wheel/ground interaction constraints. A key feature of our approach is that it incorporates appropriate physical models to cope with the task dynamics in the motion planning paradigm. The planner is based on a two-level scheme. The high level considers a simplified two-dimensional instance of the motion task and searches a subset of the configuration space of the robot in order to generate nominal subgoals through which the robot is steered. The local level solves for continuous feasible trajectories and actuator controls to move the robot between neighboring subgoals in the presence of the entire task constraints. To the best of our knowledge, this is the first implemented planner that solves for feasible trajectories to be performed by a wheeled vehicle on quite complex terrains. Simulation results are presented for the case of a six-wheeled articulated robot.