针对邮轮推舱序列自动规划问题,采用投影法建立推舱路径规划模型,并提出一种基于改进双向快速搜索随机树(Bidirectional Rapidly-Exploring Random Tree,Bi-RRT)算法嵌入的贪心算法进行邮轮推舱序列规划的方法。以大型邮轮H1508船甲板...针对邮轮推舱序列自动规划问题,采用投影法建立推舱路径规划模型,并提出一种基于改进双向快速搜索随机树(Bidirectional Rapidly-Exploring Random Tree,Bi-RRT)算法嵌入的贪心算法进行邮轮推舱序列规划的方法。以大型邮轮H1508船甲板中段区域为例,在Unity3D软件中对预制模块化舱室单元(Pre-fabricated Modular Cabin Unit,PMCU)的推舱序列规划进行仿真试验。试验结果表明,该方法可兼顾避障验证与序列规划,比传统蛇形推舱序列规划具有更高的效率。展开更多
This paper investigates the six degree-of-freedom(6DOF)relative kinodynamic motion planning problem for spacecraft close approach operations,wherein a controlled chaser spacecraft is required to approach a noncooperat...This paper investigates the six degree-of-freedom(6DOF)relative kinodynamic motion planning problem for spacecraft close approach operations,wherein a controlled chaser spacecraft is required to approach a noncooperative space target at a close range under both dynamic constraints and motion constraints.An enhanced version of the bidirectional rapidlyexploring random tree*(BiRRT*)algorithm based on flight zoning(FZ-BiRRT*)is proposed to generate safe,feasible,and nearoptimal relative motion trajectories.In the proposed algorithm,the space surrounding the space target is zoned in a spherical coordinate system based on the collision probability so that specific designs can be made for different phases of the approaching.Subsequently,based on the flight zone,dynamic constraints,and experiential knowledge,a series of modifications are made to the classic BiRRT*algorithm,and a postprocessing step is designed to accelerate convergence and promote search efficiency.Furthermore,a general regression neural network is introduced to fit a smooth and applicable final motion trajectory.Finally,the feasibility of the generated motion trajectory and the superiority of the proposed algorithm is demonstrated by means of numerical simulations.展开更多
基金supported by the Science Center Program of National Natural Science Foundation of China(62188101)the National Key Research and Development Program of China(2024YFF0504702)the SiYuan Collaborative Innovation Alliance of Artificial Intelligence Science(HTKJ2023SY502003).
摘要This paper investigates the six degree-of-freedom(6DOF)relative kinodynamic motion planning problem for spacecraft close approach operations,wherein a controlled chaser spacecraft is required to approach a noncooperative space target at a close range under both dynamic constraints and motion constraints.An enhanced version of the bidirectional rapidlyexploring random tree*(BiRRT*)algorithm based on flight zoning(FZ-BiRRT*)is proposed to generate safe,feasible,and nearoptimal relative motion trajectories.In the proposed algorithm,the space surrounding the space target is zoned in a spherical coordinate system based on the collision probability so that specific designs can be made for different phases of the approaching.Subsequently,based on the flight zone,dynamic constraints,and experiential knowledge,a series of modifications are made to the classic BiRRT*algorithm,and a postprocessing step is designed to accelerate convergence and promote search efficiency.Furthermore,a general regression neural network is introduced to fit a smooth and applicable final motion trajectory.Finally,the feasibility of the generated motion trajectory and the superiority of the proposed algorithm is demonstrated by means of numerical simulations.