A sensor guided parallel parking system for nonholonomic vehicles

Presented is a parallel parking system for nonholonomic vehicles with embedded microprocessor and ultrasonic sensors. A general four-wheel vehicle is modelled as a non-holonomic system to reflect the motion constraints applied to the vehicle. Then automated environment detection is discussed, in which a parking space is scanned and the vehicle positions itself where it is convenient for the vehicle to manoeuvre into the parking space. Based on the data acquired from ultrasonic scanning, a collision-free path is generated. The path may require a few backward and forward manoeuvres if the parking space is tight or the range of steering angle is narrow. Speed control during the manoeuvre is discussed and a scheme of speed variation is presented. The parking system has been installed on a mobile robot, and experiments have been carried out, which proves that parking spaces are detected correctly, and the path thus produced are feasible. The robot followed these paths into parking spaces safely. The system shows the potential to be integrated into automobiles.