Motion Planning for Parking a Truck and Trailer System

Abstract: Motion planning for autonomous heavy vehicles has been an intensively studied topic during the last years. In particular, real-world applicable motion planning schemes for truck-trailer systems have emerged and several continuous-curvature local planners for nonholonomic vehicles have been conceived. In this thesis, it has been studied whether it is possible to combine these results in nonholonomic motion planning to improve the efficiency of the current planning methods when planning paths for parking single-trailer truck-trailer systems in spatially constrained environments. As a part of this undertaking, a local planner for single-trailer truck-trailer systems has been developed and a comparison of the performance of different distance functions used within an RRT-planner has been made. The resulting planner was evaluated in environments with narrow passages and tight parking spaces present and the results show that the performance of the current state-of-the-art RRT-based planning methods for articulated systems can be improved in cramped environments by adding an exact local planner. Furthermore, it was discovered that heuristic lookup tables are superior to other distance functions when used within the nearest neighbor search of an RRT-based motion planning algorithm for truck-trailer systems.