基于优化的离散空间轨迹规划算法

Aiming at the problems of unsmooth path, unstable speed and long dynamic planning time in discrete space trajectory planning of unmanned vehicles, an optimization-based algorithm for discrete space trajectory planning is proposed. The space to be searched by the unmanned vehicle is decoupled into longitudinal-horizontal space and longitudinal-temporal space. In S-L space, cost function is designed for dynamic planning according to the requirements of static obstacle avoidance and path smoothing, and then the dynamic planning results are optimised using quadratic planning. In S-T space, an improved dynamic planning method is proposed to optimise the search according to road speed limits and non-reverse constraints, and the heuristic functions are introduced to speed up the search for planning endpoints, so as to reduce the algorithm's computational effort and improve operational efficiency. The simulation results show that in static and dynamic obstacle avoidance environment, the trajectory curvature of the proposed algorithm is smaller, the speed changes more smoothly, and the running time is faster. Compared with the traditional dynamic programming algorithm, the single planning time is reduced by 77.13%.