Road surface classification for extended floating car data

Abstract

An ongoing trend to connect vehicles for advanced driver assistance and driver information can be observed in the automotive industry. Whereas the channel into the car has been used for years for example by Traffic Message Compact (TMC), functions using the channel out of the car are just evolving. This paper presents a new functionality for extended floating car data (XFCD) where vehicles are used as mobile measurement probes for traffic information. It is demonstrated how vehicle data can be used to identify hazards related to the road surface. The presented algorithm classifies the current road condition using standard vehicle sensor data. The information fusion focuses on weather related events affecting the traffic safety and achieves a very good detection rate. Based on such estimations hazardous spots in the traffic network can be detected. The incar estimation of the hazard potential is presented in the context of a two-step traffic hazard recognition system. A second-level fusion combining several vehicle reports as well as additional data is performed in a central back-end server which also coordinates the provision of the valuable information to other road users as local hazard warning.