Radar Cross-Section of Potholes at Automotive Radar Frequencies

Automated detection of potholes in roads can increase driving safety for both passengers and vehicles. Potholes - especially those covered with water - are hard to detect using cameras, while vibration based detection using accelerometers cannot be used to find potholes ahead of the ego vehicle. In this paper, we study the feasibility of pothole detection using automotive radar by estimating and contrasting the pothole radar cross-section (RCS) with the backscatter from a flat road. We estimate the two-dimensional RCS of potholes from the scattered electric field simulated using finite difference time domain techniques. Our results show that the potholes give rise to significant backscatter only when the electrical properties of the pothole material are sufficiently different from the neighboring ground material. Alternatively, potholes with sloping walls forming a triangular cross-section give rise to stronger returns than those with a rectangular cross-section. The size, depth and curvature of the potholes give rise to only slight variations in the RCS. Also, the RCS from potholes filled with air and rainwater are comparable due to the low dielectric constant and conductivity of rainwater.