基于离散优化的无人驾驶汽车轨迹规划

针对双向两车道无人车行驶场景,基于离散优化的方法,提出一种新的轨迹解耦规划算法。该算法将带有时间戳的三维轨迹规划问题,解耦成分别对路径和速度规划,速度规划时引入ST图,用以描述无人车与障碍物之间的运动关系。通过分层采样的方法构建路径Lattice图搜索初始路径,以及基于多目标A*搜索算法在ST图中规划出初始速度剖面,减少算法的计算量。同时,结合优化的方法对轨迹进行优化,使轨迹收敛到全局最优解。最后,通过仿真实验,验证了该算法的有效性和可靠性。

Development of Unmanned Vehicle Trajectory Planning Based on Discrete Optimization

On basis of the discrete optimization method, a new trajectory decoupling planning algorithm is proposed on bi-directional two lanes scene for a unmanned driving vehicle. The problem of three-dimensional trajectory planning with timestamp is decoupled to the path planning and speed planning separately in this paper. First, the ST graph is introduced to describe the movement relationship between the unmanned vehicle and the obstacle in speed planning. Then, the path Lattice graph is constructed to search the initial path by use of layered sampling, and the initial velocity profile is planned in the ST graph based on the multi-object A* search algorithm, which reduces the computational consumption. Meanwhile, the trajectory is optimized to achieve the global optimal solution combining optimized methods. As a result, the simulation experiments verified the effectiveness and reliability of the proposed algorithm.