Enhanced Automated Valet Parking System Utilizing High-Definition Mapping and Loop Closure Detection

The Automated Valet Parking (AVP) system is a pivotal application of cyber-physical systems in autonomous driving, particularly for the initial and final segments of a journey. AVP system can reduce driver’s parking space cruising time and increase the capacity of parking lots. Traditional path planning approaches within AVP systems often struggle with long time for identifying target parking spaces and low efficiency in route planning, necessitating frequent adjustments. This study introduces an innovative path planning methodology that integrates high-definition mapping with loop closure detection to overcome these limitations. During parking, loop closure detection aligns the target parking space with a reference parking space from the database, streamlining the completion of parking maneuvers using established pathways. Taking parallel parking scenarios as an example, field experiments have shown that our method reduces the path length by 23.07% compared to HB-RRT, decreases the frequency of frequent adjustments by 40.0%, and improves vehicle positioning and attitude accuracy by 5.26% -10.52%. By rapidly pinpointing target spaces and calculating the most direct routes, this method enhances both parking efficiency and vehicle attitude accuracy compared to conventional methods.