Design and Laboratory Implementation of Autonomous Optimal Motion Planning for Non-Holonomic Planetary Rovers.

摘要

This work investigates the challenge of designing and implementing minimum-time trajectories for an autonomous, non-holonomic, planetary rover. The optimal trajectories were implemented at the Control and Optimization Laboratories with a TRAXXAS remote controlled vehicle modified to enable autonomous operations. These modifications include the addition of an ArduPilot controller into the architecture of the vehicle. The ArduPilot controls the inputs to the drive motor and steering servos to implement the trajectory commands generated by the trajectory optimization tool, DIDO. The challenging problem of parallel parking was used to evaluate a canonical maneuvering scenario and illustrate a procedure for motion planning that could be used for guiding a planetary rover. Three cases were evaluated with different starting points to illustrate the difficulties associated with controlling a non- holonomic vehicle. The starting points were located in front of, next to, and behind the parking space. In addition to each case, three scenarios were evaluated for complexity: no cars, two cars parked with an ideal amount of space between them, and two cars parked with minimal space between them. A VICON motion capture system was used measure the vehicle trajectory in experiments.