On-board wet road surface identification using tyre/road noise and Support Vector Machines

Changes in weather have a major influence on driving safety. On wet pavement, tyre grip is reduced and drivers must adapt their driving style accordingly. The correct operation of this adaptation mechanism depends on the perception the driver has of the asphalt status. Due to certain effects, this perception does not always correspond with reality. To improve this perception, it is essential to inform the driver about the asphalt status, efficiently and as quickly as possible. This could be achieved by installing an asphalt status detection system on the vehicle itself. The system could display asphalt status information in the vehicle’s console, allowing drivers to adapt their driving style accordingly. In this paper we propose an asphalt status classification system based on real-time acoustic analysis of tyre/road noise. The proposed system uses a practical approach that allows it to be integrated into a real vehicle. We present the system architecture used to measure the noise and the signal processing algorithms used for the classification of the asphalt state. A practical implementation of the proposed system has been developed and tested. For this preliminary prototype, only wet and dry asphalt states have been covered. Obtained wet/dry classification results have been reported, showing very high success rates.