Path planning for autonomous vehicles driving over rough terrain

This paper presents a multiresolutional architecture of planners for obstacle avoidance in outdoor mobility. The planner makes use of off-line dynamical simulations results to efficiently find paths that avoid obstacles. The trajectories are generated by sets of steering velocity commands. This approach of building the search space for the planner results in realistic cost and trajectories for the vehicle. An implementation of the planner was developed and tested in a vehicle at the National Institute of Standards and Technology (NIST) with promising results.