Wavelet‐based neural networks applied to automatic detection of road surface conditions using tire noise from vehicles.

Abstract

The detection of road surface conditions is an important process in efficient road management. In particular, in snowy seasons, prior information about the road conditions such as an icy state, helps road users or automobile drivers to obviate serious traffic accidents. This paper proposes a novel approach for automatically detecting the states of the road surface from tire noises of vehicles. The method is based on a wavelet transform analysis, artificial neural networks, and the mathematical theory of evidence. The proposed method employs the wavelet transform using multiresolution signal decomposition techniques. The proposed classification is carried out in sets of multiple neural networks using learning vector quantization networks. The outcomes of the networks are then integrated using the voting decision‐making scheme. It seems then feasible to detect passively and readily the states of the surface, i.e., as a rule of thumb, the dry, wet, snowy, and slushy state, automatically.