Direction of Movement Estimation of Cyclists with a High-Resolution Automotive Radar

Abstract

Prediction of the object movement from sensor data in the automotive sector is a widespread research and development topic. Dependent on the used sensor types, object tracking has been established over several measurement cycles. A prominent example of this is the Kalman filter. In time critical scenarios with less reaction time tracking over a number of measurement cycles is not suitable. To detect object movement within one single measurement cycle only the radar sensor is a candidate, due to the ability to measure the velocity of objects instantaneously by using the Doppler effect.A new approach to estimate the direction of movement of cyclists within one measurement cycle is introduced and explained in this paper. It is based on the approximation of the shape from a cyclist. The approximation is performed with two different methods and the solutions are compared with each other. To validate the results of the direction estimation, simulated and measured radar data are exercised.