Mpc-Based Planner for Vehicle Routing Problem with Dynamic Traffic Flows and Lights

This paper presents a Model Predictive Control (MPC)-based planner to addressing the Vehicle Routing Problem (VRP) under dynamic traffic conditions. Based on the Gaussian process, a predicting model is proposed to calculate the time of vehicles travel between different intersections with the effects of dynamic transportation. To guarantee the final path consisting of real-time decisions enable to cover all target nodes, an offline graph pruning algorithm is designed to generate targets-oriented graph list. Furthermore, within the MPC framework, the online decision making problem is optimized through the pruned graph. By framing the VRP as a time-minimized optimization problem, the MPC approach enables the vehicle to make informed routing decisions that adapt to changing conditions. Simulation results demonstrate that the proposed method achieve an average reduction of 16.12% relative to predicted fastest paths via traditional Dijstra algorithm based on static road lengths and speed limits.