Intelligent bus platoon effectively meets the high demand for public transport during rush hours. A decentralized motion planning method based on hierarchical optimization framework is proposed for intelligent bus platoon to improve flexibility and efficiency. This method realizes the optimization of the guidance trajectory in the higher layer, and the optimization of the motion trajectory in the lower layer. First, a guidance trajectories generation method is presented to decouple the platoon motion planning. The state and trajectory of the preceding bus are used as a reference to constrain the lateral and longitudinal feasible space for the current bus. Meanwhile, the quintic polynomial is adopted to generate the trajectory cluster and the optimal trajectory is selected using the multi-objective optimization. Second, the joint kinematics model of the bus is built as the prediction model. And independent model predictive control modules are utilized to optimize each guidance trajectory to make them smoother and safer. The obstacle avoidance constraints are reconstructed based on the duality theorem. Finally, simulations demonstrate the advantages of the decentralized motion planning method under different scenarios, and results demonstrate our method improves the flexibility and consistency of the platoon.

智能公交编排有效满足高峰期公交高需求。为提高智能公交队列的灵活性和效率，提出了一种基于分层优化框架的分散式运动规划方法。该方法实现了上层制导轨迹的优化，下层运动轨迹的优化。首先，提出了一种引导轨迹生成方法来解耦排运动规划。以前车状态和轨迹为参考，约束本车横向和纵向可行空间。同时，采用五次多项式生成轨迹聚类，并使用多目标优化选择最优轨迹。第二，建立客车关节运动学模型作为预测模型。并利用独立的模型预测控制模块优化每条制导轨迹，使其更平稳、更安全。基于对偶定理重构避障约束。最后，仿真证明了分散运动规划方法在不同场景下的优势，结果表明我们的方法提高了排的灵活性和一致性。