An integrated approach that optimizes traffic signals and vehicle trajectory at urban intersections using Visible Light Communication (VLC) is proposed. A Connected Vehicle (CV) platoon approaches a signalized intersection, and downstream CVs queue before the stop lines. Light is used to communicate information between CVs and the infrastructure using streetlamps, intersection signals, and headlights. Interaction with traffic is coordinated by an intersection manager. Integrated control is flexible and adaptive to traffic demands since different traffic movements are incorporated during multiple signal phases. As part of the simulation process, an open-source urban mobility simulator SUMO creates the desired scenarios and generates different urban traffic flows. VLC queue/request/response mechanisms and temporal/space relative pose concepts are used. Using sequence state durations, phase diagrams, and average speed measurements, the system dynamically controls traffic flows at intersections using a Deep Reinforcement Learning (DRL) algorithm, minimizing rush hour bottlenecks, through joint Vehicle-to-Vehicle (V2V) and Vehicle-to-Infrastructure (V2I) communications. Comparisons with trajectory optimization and signal optimization demonstrate the benefits on throughput, delay, and vehicle stops, and reveal the optimal patterns for signals and trajectory.
|