Parking space detection from a radar based target list

The detection of an unoccupied parking spaces is a requirement of autonomous parking systems. This paper presents a radar based real-time algorithm, which detects, classifies, and evaluates parking spaces in a vehicle's immediate vicinity. In contrast to frequently used gridmap methods, our approach processes the radar data in a specially designed target list of 2D vectors. Consequently, quantization errors are avoided and computational burdens reduced. Experiments show that the proposed algorithm is suited for both parallel and perpendicular parking spaces in urban scenes.