密集障碍物环境下移动机器人全局路径规划

Global Moving Robot Path Planning under the Environment of Dense Obstacles

针对粒子群优化算法用于障碍物密集分布环境下机器人全局路径规划存在的早熟、效率低等问题,提出了一种基于障碍物顶点信息搜索的双层(底层和顶层)粒子群优化算法。首先,循环运行若干次底层算法,快速获取若干条无碰路径,确定全局最优解的大致位置,并上传所得路径到顶层;顶层种群接受下层信息后,接着,进行局部精细搜索,以获取问题的最优解。同时,定义了基于障碍物顶点信息的脱障算子,对粒子的全局极值点进行脱障操作,以保证路径的无碰性且加快寻优效率。最后,仿真验证了该方法的有效性。

Aimed at the disadvantages such as premature convergence and low efficiency when apply particle swarm optimization to robot path planning with dense obstacles, the paper proposes a two-layer （bottom and top）particle swarm optimization. based on the vertex information of obstacles. First, the bottom PSO is run to obtain a number of collision-free paths quickly, and to determine approximate position of the global optimal solution, then upload these paths to the top level. After receiving these information from the bottom PSO, the top PSO implements a locally fine search in order to locate the global optimal solution. Meanwhile, an off-barrier operator based on vertices information of obstacle is defined to modify the global best position of particle, it collide still with obstacles, in order to ensure the feasibility of path and speed up the optimization speed. Finally, the simulation verifies the effectiveness of the proposed method.