Simultaneous Odometry, Mapping and Object Tracking with a Compact Automotive Radar

This paper presents the first results of an experimental study about the possibility to implement basic vehicle odometry, road structure mapping and moving object tracking functionalities by combining a compact automotive radar sensor with appropriate digital signal processing algorithms. In the context of the study, radar raw data sets for real-world road scenarios, including static and moving objects, are first collected from a vehicle-mounted, single-chip 77 GHz automotive radar transceiver and stored on a hard disk. The acquired raw data sets are then processed offline by an algorithm which attempts to detect stationary and moving road objects, estimate odometry data for the radar vehicle and for the detected moving objects and build a map of the detected road structures. Considerations on the applicability of the presented approach and possible extensions of the research work are discussed in conclusion.