A Safe Motion Planning and Reliable Control Framework for Autonomous Vehicles

Abstract

Accurate trajectory tracking is unrealistic in real-world scenarios, however, which is commonly assumed to facilitate motion planning algorithm design. In this paper, a safe and reliable motion planning and control framework is proposed to handle the tracking errors caused by inaccurate tracking by coordinating the motion planning layer and controller. Specifically, motion space is divided into safe regions and risky regions by designing the movement restraint size dependent on tracking error to construct the repulsive potential field. The collision-free waypoint set can then be obtained by combining global search and the proposed waypoint set filtering method. The planned trajectory is fitted by an optimization-based approach which minimizes the acceleration of the reference trajectory. Then, the planned trajectory is checked and modified by the designed anti-collision modification to ensure safety. Using invertible transformation and adaptive compensation allows the transient trajectory tracking errors to be limited within the designed region even with actuator faults. Because tracking error is considered and margined at the planning level, safety and reliability can be guaranteed by the coordination between the planning and control levels under inaccurate tracking and actuator faults. The advantages and effectiveness of the proposed motion planning and control method are verified by simulation and experimental results.