Urban Driving Powered by Radar Based Deep Learning: A trained network to identify parked cars

摘要

For automated driving in urban areas and for valet parking radar sensors provide a robust detection. With modern deep learning methods the car can understand the environment based on the fusion of the different sensor signals and identify landmarks for self-localization sensor individually and fused. Within this the classification of parked cars is important for a reliable and safe localization. The paper describes the challenges for urban driving, the use case within that of valet parking and explains the result of a trained network, which is able to identify parked cars.