Vehicle-to-Infrastructure based Algorithms for Traffic Light Detection, Red Light Violation, and Wrong-Way Entry Applications

Major injuries and fatalities in the United States (US) frequently occur as a result of drivers attempting to run red lights at the intersections and driving in the wrong-way on access ramps or against the legal direction of the traffic leading to red light violations (RLVs) and wrong-way entry (WWE) events respectively. But, these incidents can be detected earlier using connected vehicle (CV) technology which may then be reported to the drivers through a human-machine interface. In this work, we propose algorithms to detect the approaching traffic light status, red light running (RLR), and wrong-way driving (WWD) events using Basic Safety Messages (BSMs) obtained through vehicle-to-infrastructure (V2I) communication. To detect the accurate and applicable light, we assume that road-side units (RSUs) are installed on each traffic light in such a way its heading direction is opposite to the legal flow of traffic. After the approaching light status is detected, we check if the vehicle is going to be in an RLR zone. Similarly for WWD, we assume the same except that the RSUs are installed on street poles. After the detection, we display the traffic light status, red light violation warning (RLVW), and wrong-way entry warning (WWEW) to the drivers in real time. To validate the proposed algorithms, the CARLA simulator is used to conduct experiments for various scenarios. Also, to support connectivity between vehicles and infrastructure, we added the V2I ability to CARLA.