Evolutionary Algorithm For Multi-agent Robot Path Planning

Multi-agent Robot System(MARS) with single robot system, which can not match the many advantages of robotics is gradually becoming an important research direction. Path planning is one of the most fundamental and most important tasks in Multi-Agent Robots System. And it shows the robot's exchanging ability with around environment, therefore, Path Planning Research for Multi-Agent Robot in complex environments is important and practical.In this thesis, path planning problem for Multi-Agent robots in complex environments is comprehensively studied and two methods are presented to deal with the problem. The first method is based on the Differential Evolution. And the second one is suggested by Co-Evolution method. The result of computer simulation shows that the two methods has a good convergence, robustness and high planning speed and tracking accuracy.The path planning method based on Differential Evolutionary(DE) algorithm is a new type of evolutionary approach. DE will be introduced to the multi-robot path planning, it possessed a parallel, ease of use, robustness and good strong global optimization the ability to overcome the traditional characteristics of the genetic algorithm path planning evolution inefficient, complex genetic operation defects. Meanwhile, an effective solution to the artificial potential field path planning capacity in premature convergence and deadlock is presented in DE.The Co-Evolution(CE) algorithms for multi-robot path planning is used a thinking of multiple robot co-evolution to optimize the overall planning path. The shortcomings of slow path planning in the traditional evolutionary computation approach will be overcomed. new algorithm accelerated multi-robot path planning, the results of the study to maintain diversity and avoid a local optimum in path planning.