Road Surface detection Using FMCW 77GHz Automotive RADAR using MFCC

Abstract

Road accidents can be avoided to a greater extent by identifying the kind of road surface and thereby alerting the driver in advance. This helps the driver to immediately take decision on the speed of the vehicle. This paper puts forth a technique for automatic identification of road surface type using FMCW 77GHz Automotive RADAR. It is based on the analysis of the backscattered signals of the RADAR using Mel Frequency Cepstral Coefficient and building a Classifier model for the training data. The novelty of this technique is the data integration, feature extraction followed by the road surface type recognition using classifiers like KNN (K-Nearest Neighbors), DT(decision tree) and SVM (Support Vector Machine). This technique identifies five road surface types i.e. dry concrete, grassy, slush surface dry asphalt and sand. The experimental results show accuracy above 97% using SVM classifier. Also the response time of the system is calculated in real time, for various speeds of the vehicle. It is observed the driver is alerted around 19.5m before the point of observation is reached, when the vehicle is moving at the speed of 40kmphr.