Detection and Simple Tracking Method for ADAS using a mmWave Radar

Advanced Driver Assistance Systems (ADAS) are becoming reality now. Several researchers are attempting to innovate new methods for vehicle collision avoidance, automated parking, self-driven cars, etc. Towards this area, this paper proposes a mechanism for the detection of targets and a simple tracking method for tracking the same. An experiment was carried out using 77 GHz mmWave Radar that was mounted on a car. The Radar was used to track the moving vehicles as well as stationary objects in front of the ego vehicle. The Radar raw data was recorded in real time while the ego vehicle was on the move. The raw data was then processed to obtain Range-Doppler Map (RD Map). Further Range vs Speed map was obtained from RD-map. 2D Constant False Alarm Rate (CFAR) algorithm was performed on Range vs Speed map for detection of targets on each frame. Further, CFAR detection map of each processed frame was cumulatively added to form a single Detection and Tracking Map (DTM). DTM shows the detection of new targets as well as the history of older targets as tracks over a period. Older targets are gradually deleted from the DTM by applying the deletion rule. The experiment was conducted with the ego vehicle on the road, moving with a speed of around 30 kmph. The Radar raw data obtained with this setup and the proposed algorithms can be used as an input for development of advanced algorithms for collision avoidance, automated car parking and even for self-driven cars.