基于主从结构的多水下机器人协同路径规划  

Multi-AUV Cooperative Path Planning Based on Master-Slave Structure

作　　者：李东正[1] 郝燕玲[1] 张振兴[1] 

机构地区：[1]哈尔滨工程大学自动化学院,黑龙江哈尔滨150001

出　　处：《计算机仿真》2015年第1期382-387,共6页Computer Simulation

摘　　要：关于多水下机器人协同路径规划问题,是多水下机器人协同控制的重要研究内容之一,是一种典型的含多个约束条件的组合优化问题。针对多机器人协同路径规划因约束条件多导致算法复杂度高、耗时、求解困难等问题,提出了一种主从结构的并行多水下机器人协同路径规划算法。进化过程的每一代,子层结构应用粒子群并行算法,生成各架机器人当前的最优路径,同时,主层结构应用微分进化算法实时给出当前考虑机器人与障碍物、机器人与机器人之间避碰情况下,总系统运行时间最短的路径组合方案。上述结构将多约束分解到不同层面,有效地降低了单层结构因过多的约束条件计算时间过长以及不易实现等困难。仿真结果表明,上述算法不仅能在静态环境下生成可行的、优化的组合路径,而且在当障碍物随时间随机移动的动态环境下,也表现出可行的、良好的效果,为求解多水下机器人协同路径规划问题提供了一个高效的解决方案。Multi-AUV cooperative path planning, a combinational optimum problem with constraints, is one of the most important research fields of multi-AUV cooperative control. Because of above reasons, a parallel programming algorithm based on master and slave structure was proposed in this paper. For slave level, PSO was used to generate a feasible path for each robot, while DE was used in the master level, considering the collision, to generate low time -cost combinational optimum paths for system. The simulation results show that feasible and optimal paths in stationa- ry environment are generated by the proposed master-slave algorithm, moreover, the algorithm performs well on the application of dynamic path planning when the obstacles cruise randomly. What＇ s more, the performance of multi- AUV path planning is improved and the theory of collision and cooperation is enriched.

关 键 词：多水下机器人 主从结构 协调路径规划 避碰 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]