Information-Centric Approach to Autonomous Trajectory Planning Utilizing Optimal Control Techniques

摘要

This work introduces a new information-centric pseudospectral optimal control-based algorithm for autonomous trajectory planning and control of unmanned ground vehicles with real-time information updates. It begins with a comprehensive study and comparison of the various path planning methods currently in use. It then provides an analysis of the optimal control method, including vehicle and obstacle modeling techniques, several different problem formulations, and a number of important insights on unmanned ground vehicle motion planning. The new algorithm is then utilized on a collection of motion planning scenarios with varying levels of information; the performance of the planner and the solution accuracies under these varying levels of information are studied for both single and multi-vehicle scenarios. The multi-vehicle scenarios compare and contrast centralized, decentralized, decoupled, coordinated, cooperative, and prioritized control methods. Finally, the versatility of the planner (and the optimal control technique) is demonstrated, as it is used as both a path follower and trajectory planner in a collection of scenarios, including multi-vehicle formations and sector keeping.