Real-Time Cooperative Trajectory Planning for Multiple CAVs at Unstructured Intersections: A Computational Optimal Control Approach

This paper is about real-time cooperative trajectory planning for automated vehicles at lane-free, unsignalized, and unstructured intersections. Such scenarios are challenging due to their irregular boundaries, narrow space, and dense heavy-vehicle traffic, creating congestion that traditional methods struggle to resolve. To address these issues, we propose a computational optimal control framework. First, a zero-DOF simulation-based planner is proposed to generate kinematics-feasible reference trajectories rapidly, avoiding the computational burdens of conventional search-based or mixed-integer programming-based methods. Second, based on the derived reference trajectories, we construct linear and scalable collision-avoidance constraints through safe travel corridors and relative safe travel corridors, largely simplifying the nominally high-dimensional collision-avoidance constraints. Third, a lightweight iterative optimization strategy is applied to resolve the underlying conflicts between corridor constraints and vehicle kinematics, ensuring robust convergence to near-optimal cooperative trajectories. Extensive simulation results verify that the proposed framework improves intersection throughput and computational robustness.