Environment reconstruction and trajectory planning for automated vehicles driving through signal intersection

Because signalized intersections are complex driving environments, driving behavior improvement research at signalized intersections has become a hotpot in the field of intelligent transportation and vehicle control. We aim to improve the efficiency, economy, and comfort of autonomous vehicles (AVs) driving through signalized intersections. We construct a driving risk field to describe the dynamic and static environment of a signalized intersection simultaneously by combining molecular dynamics and field theory, in which the dynamic potential expresses both the attraction and repulsion in all directions between adjacent vehicles. Then, an objective function for AV trajectory optimization is established by considering driving efficiency, energy consumption, and comfort. The results indicate that compared with the adaptive cruise control (ACC) model, our model can reduce the average travel time by more than 10% and improve the energy consumption and comfort by 18.9% and 16.0%, respectively at the highest among the three driving directions. In addition to optimizing AV driving at signalized intersections, this research can also be used to simulate driving behaviors, optimize signal timing schemes, and design channelization schemes by expressing the risk at any point in the signalized intersection at any time, as well as the dynamic change in the spatial distribution of risks over time. Our work is even conducive to the research of adaptive signal control and dynamic markings and provides a framework for researching intelligent vehicle–infrastructure cooperative systems.