An Autonomous Driving Model Integrated with BEV-V2X Perception, Fusion Prediction of Motion and Occupancy, and Driving Planning, in Complex Traffic Intersections (2312.05104v3)

Abstract: The comprehensiveness of vehicle-to-everything (V2X) recognition enriches and holistically shapes the global Birds-Eye-View (BEV) perception, incorporating rich semantics and integrating driving scene information, thereby serving features of vehicle state prediction, decision-making and driving planning. Utilizing V2X message sets to form BEV map proves to be an effective perception method for connected and automated vehicles (CAVs). Specifically, Map Msg. (MAP), Signal Phase And Timing (SPAT) and Roadside Information (RSI) contributes to the achievement of road connectivity, synchronized traffic signal navigation and obstacle warning. Moreover, harnessing time-sequential Basic Safety Msg. (BSM) data from multiple vehicles allows for the real-time perception and future state prediction. Therefore, this paper develops a comprehensive autonomous driving model that relies on BEV-V2X perception, Interacting Multiple model Unscented Kalman Filter (IMM-UKF)-based fusion prediction, and deep reinforcement learning (DRL)-based decision making and planning. We integrated them into a DRL environment to develop an optimal set of unified driving behaviors that encompass obstacle avoidance, lane changes, overtaking, turning maneuver, and synchronized traffic signal navigation. Consequently, a complex traffic intersection scenario was simulated, and the well-trained model was applied for driving planning. The observed driving behavior closely resembled that of an experienced driver, exhibiting anticipatory actions and revealing notable operational highlights of driving policy.