Estimating friction from surface roughness: is a PSD enough?
The most common road surface characterization metric used in friction models is power spectral density (PSD). However, PSDs tell us little about how roughness is distributed locally on the surface and may misinterpret the frequency signature of edge fall-offs as high-frequency roughness. A new method for surface-scan processing is presented, which uses the frequency content, as well as the height and gradient distributions of the roughness. This method results in the generation of realistic virtual surfaces which, combined with a novel approach for friction estimation, highlight the importance of accurate road representation.