Application Of A Stream-Based SAR-Backprojection Approach For Automotive Environment Perception

This paper presents the vehicle integration of a synthetic aperture radar (SAR) algorithm to produce SAR images for automotive scenarios while driving. The focus of this work is on two parts. At the beginning, the hardware setup embedded in the test vehicle is presented as well as the sensor synchronization to provide SAR raw data. Subsequently, a pixel-based parallelization concept for stream-based SAR image calculation on a GPU is introduced. Our experimental results prove that the complete SAR imaging process can be integrated into a vehicle to produce online highly resolved amplitude maps for automated driving applications.