Autonomous Reverse Parking System Based on Robust Path Generation and Improved Sliding Mode Control

摘要

Some commercial vehicle models have been equipped with semiautonomous parking systems to a certain extent. However, gaps to fully automated solutions still exist, and cost considerations further constrain their acceptance among consumers. This paper proposes a low-cost vision-based approach to a fully self-reverse parking system. It consists of four key modules: a novel path-planning module ensures that a feasible path is available under any initial poses, which frees human intervention completely; a modified sliding mode controller on the steering wheel is designed for path following; image processing with Kalman state prediction provides consistent and real-time estimation on the vehicle pose; and a robust overall control scheme ensures that the vehicle can accurately park along the slot center line without intrusion into adjacent slots. Experimental results based on 216 on-field tests under different illumination conditions showed that the proposed system was able to accurately and consistently park the vehicle in all cases with a 4.71-cm RMS offset distance from the center line and a 1.24 RMS orientation deviation. With its easy setup and excellent performance, this system can be practically and robustly implemented to existing vehicles with minimal additional cost.