Research Of Multiple Mobile Robots Path Planning Under Dynamic Environment

Path planning for multiple mobile robots is a challenging problem in robotics andis core content of robot navigation. Path planning for multiple mobile robots underdynamic environment refers to find an optimal and collision-free path for each robotfrom the start point to the goal point in a workspace with static and dynamic obstacles.So far, most of existing researches on the multi-robot path planning problem are underknown environment without dynamic obstacles. Multiple mobile robots path planningunder dynamic environment is still a difficult problem which still has to be resolved.This paper aims to solve multiple mobile robots path planning problem under dynamicenvironment. And then presents a multiple robots distributed rolling double-level pathplanning algorithm based on the research results of single mobile robot path planningwhich uses the hybrid methods of global path planning and local collision avoidanceplanning. This paper mainly includes the following contents:1. Research the single mobile robot path planning in global environment, andpropose an improved genetic algorithm based on the conventional genetic algorithm. Inthe progress of algorithm design, more accurate fitness function is employed. Selection,crossover and mutation genetic operator are improved and a new genetic operator isproposed. The improved genetic algorithm overcomes the defects of conventionalgenetic algorithm, such as falling easily into local optimum and slow convergence speed,and is superior to classic ant colony algorithm at the performance of shortest path lengthand time.2. Research the single mobile robot path planning under dynamic and complicatedenvironment. Employing rolling window principle, a double-level path planningalgorithm based on the combination of global path planning and local rolling collisionavoidance planning is presented. In the process of global path planning, an improved antcolony algorithm is proposed. In the process of local rolling collision avoidanceplanning due to the uncertainty of the moving direction of dynamic obstacles, twocollision prediction plans are presented and two collision avoidance strategies areproposed accordingly. And because the proposed double-level planning algorithm is unable to adapt to the changing environment, the Follow_wall action is used in thesecond level planning to improve the algorithm. Simulation experiment demonstratesthe effectiveness of the proposed double-level planning algorithm and the superiority ofthe improved ant colony algorithm.3. Research the multiple mobile robots path planning in dynamic environment.Based on the research results of single mobile robot path planning, multiple robotsdistributed rolling double-level planning algorithm is put forward. And aims at thepossible path conflicts between mobile robots, a mixed coordination strategy isproposed. Then a simulation that multiple mobile robots go through narrow passage isdesigned, which indicates the effectiveness of the proposed algorithm.