Identify method of road surface conditions based on the statistical pattern recognition technique

As the Internet of Things developed quickly, the vehicle-road collaboration system will be the key part of intelligent transportation system. The technology also creates new opportunities for enhancing vehicle safety. Because travelled on the road, it is significant for the vehicles to know the surroundings. To meet the requirements of the highway real-time safety prediction system under Internet of Things, a road surface condition identify method based on statistical pattern recognition technique is proposed. The method can be improved in common driving situation, no need for special scenario. In this paper, a sensitive index extracted and the probability distribution of the index is calculated off-line. On-line the Bayesian rules are used to update the real-time probability of the index and the posterior probability. Finally the maximum-likelihood estimation for the probability of each road condition is computed and identified the type of road condition. The result is correct and could be reference for future research.