Pursuit Model Development and Calibration Methodology for Automated Parking Function

Automated Valet Parking is foreseen as a highly automated function to be implemented on large scale in mass production in the near future. For such function, The authors have developed an innovative system concept using both real time kinematic sensing system and infrastructure IoT (Internet of things) camera to perform AVP function. As the driver activates the AVP function, the infrastructure identifies available parking spots by processing in real time camera frames and as the path is computed, the vehicle executes automated parking maneuvers. This paper introduces the system scheme and system engineering approach as well as key technology bricks to realize AVP function and proves the effectiveness and accuracy of the system through extensive vehicle testing. From system definition to final prototype validation, through Motion control algorithm description or deep learning algorithm tuning method, this paper covers a large spectrum of the engineering domains required to implement such function. Finally a specific Pursuit Model development and calibration approach is elicited for the motion control function and proven in use by AVL Vehicle Prototype platform.