Millimeter-Wave AWR1642 RADAR for Obstacle Detection: Autonomous Vehicles

Abstract

To know the nature of the surrounding is very crucial, and the profile of the terrain geometry must be known to decide the motion path in autonomous cars. The obstacles must be detected along with their coordinates with the vehicle as origin (reference) and relative velocity to determine a collision free path. Most autonomous vehicle manufacturers depend extensively on LIDAR such as Uber and Google, whereas Tesla uses RADAR as a primary sensor. RADAR is relatively less expensive and works equally well in all weather conditions such as fog, rain, snow, and dust. It can also determine relative traffic speed using Doppler frequency shift which is not possible in case of LIDAR. Hence, RADAR is a viable solution for autonomous cars considering above parameters, and it provides good results when used with secondary sensors such as ultrasonic and camera. This paper proposes a methodology for testing RADAR sensor AWR1642 in various configurations with respect to distance and angle of coverage. The testing of the above RADAR is carried out on real vehicle. The results obtained verify that RADAR provides accurate results within 50 m range for medium-sized objects with appropriate tuning of parameters such as best range and best range resolution.