Applying behavior recognition in road detection using vehicle sensor networks

Transportation related environment and social problems become very imminent with rapid urban growth. Road detection aims to detect road anomalous events that endanger vehicle driving. Road detection using Vehicle Sensor Networks (VSNs) is a new method with many advantages, as vehicles become more and more popular nowadays. Data collected by vehicle sensor network is related with vehicle behaviors such as trajectories and driving models, etc. In this paper, we propose a scheme to apply Behavior Recognition in Road Detection (BRRD) by using VSNs on road. A dynamic Bayesian network model is used to infer the road events according to vehicle behaviors recognized by sensing data. Based on it, we propose group detection by utilizing collaborative filtering to improve detection performance. Our experiments on testbed of 3 remote controllable vehicles simulate transportation and few road condition scenarios. The results show that our solution provides effective detection with more than 91% accuracy, minor false positive and false negative.