Segmentation of a road from a vehicle-mounted radar and accuracy of the estimation

This work treats a problem relevant to automatic road following, collision avoidance, and maneuver control. A millimeter-wave radar is placed on the front of the vehicle and an image of the radar backscatter of the terrain is acquired in polar coordinates. This image contains both road and off-road scatter components which must be segmented from each other. Geometric constraints on the road edges are naturally formulated in the Cartesian cartographic domain. However, to use the recorded data directly, we apply the road constraints in polar coordinates. Moreover, our estimator is based on a criterion which is largely insensitive to the off-road scatter. Numerical results are given for real data and illustrate the accuracy and robustness of our approach. Finally we derive the Fisher information matrix for evaluation of the achievable accuracy of segmentation for our model via the Cramèr Rao bound.