Automotive Velocity Sensing Using Millimeter-Wave Interferometric Radar

In this article, we demonstrate the use of interferometric radar for high-accuracy measurement of the ground speed of a moving vehicle. Recent work has shown the capability of radar systems using interferometric antenna apertures to directly measure the angular velocity of objects. In this work, we apply this new measurement to the automotive application of velocity estimation. We start by formulating the interferometric response from a ground surface, which is modeled as a summation of independent point scatterers. Next, we present two velocity estimation methods: a Bayesian method that uses a signal model and a neural network-based approach. We then describe a 40-GHz interferometric radar system that we have developed and present on-vehicle measurements. Finally, we present the results of the two estimators and compare these results with the theoretical accuracy of the system. We show that the Bayesian and neural net estimators achieved a root-mean-square velocity error of 0.403 and 0.183 m/s, respectively.