Infrastructure-cooperative algorithm for effective intersection collision avoidance

To guarantee the road safety by avoiding collisions at the intersections is one of the major tasks of intelligent transportation systems (ITSs), which contributes to the minimal fatalities and property loss in crashes. This paper proposes an effective algorithm for infrastructure-cooperative intersection accident pre-warning system with the aid of vehicular communications. The proposed algorithm realizes accurate and efficient collision avoidances through five steps, i.e., defining variable, reasoning the vehicles evolution state, verifying safe driving behavior, assessing risk, and making decision. The critical factors are theoretically analyzed, and a vehicle state evolution model based on the Dynamic Bayesian Networks (DBNs) is established. The efficient risk assessment method based on identifying the dangerous driving behavior at intersection and different collision avoidance strategies are proposed according to the actual situation. Finally, extensive simulations are carried out to verify the performance of the proposal, and simulation results show that the proposed algorithm can effectively detect risk and accurately migrate the collision.