An Eco-Driving Method with Queue Length Estimation for Connected Vehicles

Aiming at speed planning problems for connected vehicles traveling through multiple traffic signals under a dynamic traffic environment, an eco-driving method was proposed based on real-time queue length estimation. Firstly, a radial basis function neural network was constructed and trained to estimate queue length at intersection. Then, in the frame of optimal control, the traffic queuing was mathematically modeled together with traffic signals to formulate a speed profile optimization problem. Finally, the proposed decoupling transformation method was used to calculate a reference speed profile efficiently. Simulation results reveal that the proposed method can provide smoother actual speed profiles and save more than 40% energy compared with the traditional eco-driving method without considering the traffic queuing.