Spatiotemporal Speed Planning for Multi-Vehicle Cooperative Driving With Dynamic Following Relationships
2687-7813
Abstract
In Autonomous Valet Parking scenarios, multiple vehicles operate cooperatively in low-speed environments where the leading-vehicle constraint may be dynamically reconstructed due to path transitions. Under such conditions, conventional speed planning methods struggle to maintain stable following behavior and temporal consistency. To address this issue, a spatiotemporal modeling-based multi-vehicle cooperative speed planning and optimization method is proposed. First, assuming that the geometric paths of vehicles are known, a continuous and predictable speed evolution model is constructed for the leading vehicle. The corresponding trajectory is projected into the Space-Time graph to form dynamic obstacles, enabling a unified representation of temporal relationships and collision avoidance constraints among multiple vehicles. On this basis, a constraint-guided Hybrid A<inline-formula> <tex-math notation="LaTeX">${}^{\ast }$ </tex-math></inline-formula> search strategy is introduced to generate feasible speed trajectories satisfying vehicle kinematic constraints in the space-time graph. Furthermore, considering the problem of speed mutation, a speed optimization and verification mechanism is designed, and the feasibility and safety of the speed curve are evaluated through a multi-index verification mechanism. Simulation results demonstrate that the proposed method can stably generate continuous and feasible speed trajectories under several representative parking scenarios. The method improves motion safety and comfort while enhancing the overall efficiency of multi-vehicle cooperative driving in parking environments.