A Spatial-Domain Coordinated Control Method for CAVs at Unsignalized Intersections Considering Motion Uncertainty (2412.04290v2)

Abstract: Coordinated control of connected and automated vehicles (CAVs) emerges as a promising technology to improve traffic safety, efficiency, and sustainability. Meanwhile, mixed traffic, where CAVs coexist with conventional human-driven vehicles (HDVs), represents an upcoming and necessary stage in the development of intelligent transportation systems. Considering the motion uncertainty of HDVs, this paper proposes a coordinated control method for trajectory planning of CAVs at an unsignalized intersection in mixed traffic. By sampling in distance and using an exact change of variables, the coordinated control problem is formulated in the spatial domain as a nonlinear program, thereby allowing for unified linear collision avoidance constraints to handle vehicle crossing, following, merging, and diverging conflicts. The motion uncertainty of HDVs is decoupled and modeled as path uncertainty and speed uncertainty, whereby the robustness of collision avoidance is ensured in both spatial and temporal dimensions. The prediction deviation for HDVs is compensated by receding horizon optimization, and a real-time iteration (RTI) scheme is developed to improve computational efficiency. Simulation case studies are conducted to validate the efficacy, robustness, and potential for real-time application of the proposed methods. The results show that the proposed control scheme provides collision-free and smooth trajectories with state and control constraints satisfied. Compared with the converged baseline, the RTI scheme reduces the computation time by orders of magnitude, and the solution deviation is less than 2.3%, demonstrating a favorable trade-off between computational effort and optimality.