一种针对月球飞行探测器的运动基元规划方法

A fast optimal kinodynamic motion planning method for LFV is proposed,taking into account the absence of prior map and unknown obstacle distribution. The method is based on motion primitives,heuristics and terminal shooting. Firstly,a comparison is made between primitive generation models based on constant thrust and constant acceleration. To reduce the time of constraint checking,it is suggested that primitives should be designed in the acceleration control space. In order to address the challenge of generating primitive families fastly due to the time-varying non-convex acceleration control space,a deterministic uniform sampling strategy based on dispersion and a screening method of effective sampling points are designed. This sampling strategy minimizes the number of sampling points while improving coverage of primitives within the reachable domain. By establishing extremum expressions for position and velocity of primitives,a filtering method converts the entire primitive into a single state,thereby reducing the number of constraint checks required. Finally,a heuristic method suitable for LFV is designed to reduce algorithm running time,and a hierarchical terminal shooting strategy is developed to increase success rate in connecting target state with trajectory tree. Simulation results demonstrate that this proposed method effectively balances optimality with fast computation while ensuring flight safety for LFV.