STdi4DMPC: Distributed Model Predictive Control for Connected and Automated Truck Platoon With Mixed Traffic Flow Based on Spatiotemporal Trajectory Prediction

Trucking transportation has brought great convenience to social life, but it also faces many challenges. The rapid development of autonomous driving technology has the potential to alleviate some bottleneck issues and lead to new changes in truck transportation. Ensuring the high-quality operation of connected and automated trucks (CAT) and connected and automated trucks platoon (CATP) in mixed traffic flow has sparked a wave of research. With the goal of achieving efficient and stable longitudinal control of CATP in mixed traffic flow, this paper develops a novel data-driven longitudinal control framework for CATP based on trajectory prediction (STdi4DMPC). Firstly, a spatiotemporal dynamic attention model for trajectory prediction using a Transformer style architecture that integrates driving intentions (STdi) is constructed, capturing trajectory features of different dimensions and dependencies at different scales. Additionally, processing trajectories using empirical mode decomposition and wavelet denoising. Furthermore, a CATP longitudinal controller that conforms to the truck dynamics constraints based on distributed model predicted control (DMPC) is developed. Finally, the reconstituted noise-reduced trajectory of human-driven vehicle (HDV) is employed as a reference trajectory for DMPC, thereby achieving the integration of trajectory prediction model and longitudinal control algorithms. The simulation results based on the CitySim dataset are conducted and results demonstrate the superiority of the proposed longitudinal control strategy.