A Hidden Markov Model based driver intention prediction system

Awareness of other vehicle´s intention may help human drivers or autonomous vehicles judge the risk and avoid traffic accidents. This paper proposed an approach to predicting driver´s intentions using Hidden Markov Model (HMM) which is able to access the control and the state of the vehicle. The driver performs maneuvers including stop/non-stop, change lane left/right and turn left/right in a simulator in both highway and urban environments. Moreover, the structure of the road (curved road) is also taken into account for classification. Experiments were conducted with different input sets (steering wheel data with and without vehicle state data) to compare the system performance.