Vehicle Behavior Analysis for Uneven Road Surface Detection

Although on-road vehicle detection and tracking is a well-researched area, vehicle behavior analysis is less addressed and still in a nascent stage. In our study, after successful vehicle detection and tracking, vehicle behavior is modeled using two dimensional spatio-temporal trajectories for uneven road surface detection. Uneven road surfaces are identified from specific motion patterns which are described using a Hidden Markov Model based trajectory classification model. We claim that capturing the sudden change in vehicle´s vertical motion pattern can suggest an anomaly, like a speed bump or a pothole on road.