An Efficient On-Ramp Merging Strategy for Connected and Automated Vehicles in Multi-Lane Traffic

On-ramp merging scenario has a great impact on traffic efficiency and fuel economy. At present, most research on-ramp merging focuses on the optimization of merging sequence in the single main lane scenario, which fails to give full play to the capacity of multi-lane roads. To overcome this problem, an efficient on-ramp merging strategy (ORMS) is proposed to coordinate vehicle merging in multi-lane traffic. First, we built a model of the unevenness of traffic flow between lanes. Based on this model, we established a lane selection model by reinforcement learning for the coordination of vehicles in multi-lane traffic. Before vehicles enter the merging zone, the decision of lane selection is made by analyzing the unevenness of traffic flow between lanes to relieve local congestion in the outside lane that may be caused by ramp vehicle inflow. Then, we adopted a vehicle motion planning algorithm based on the time-energy optimal control, so that all vehicles travel according to the optimal trajectory to reach the merging zone. The simulation results show that the traffic efficiency and fuel economy of the proposed on-ramp merging strategy are significantly improved compared with the existing optimal control algorithm.