Multi-robot Co-evolution Path Planning Based On Perception Community Environment

Multi-robot system has distributed characteristics in space, time, and functions with strong robustness and reliability. Design, application control, and others of multi-robot systems has become a hot research topic. For some complex tasks, particularly concurrent operation missions, which is difficult to perform for single robot, the coordination and cooperation is important among multi-robots. Path planning is one of the important research directions for multi-robot systems. It is necessary to collaborate and coordinate for multiple robots to complete path planning in robot soccers, workshop production line and so on. Under unknown environment and community environment for multi-robot systems, using genetic algorithm, this paper-focuses on cooperation within the sub-populations and sub-populations of evolutionary through the modified algorithm, and put forward some improved collaborative evolutionary path planning strategies here.Firstly, The research status on the multi-robot path planning are reviewed in this paper according to graph theory based environment modeling, RRT based environment modeling, evolutionary algorithm, swarm intelligence, fuzzy logic algorithm and behavior-based method, respectively. Moreover, the related issues about the task allocation and obstacle avoidance of multi-robot path planning are analyzed. Then, the challenges and the trend of multi-robot path planning are given from the environmental perception information sharing and fusion, collaboration and coordination optimization mechanism and formation path planning of multi-robot system.Secondly, a path planning strategy based on co-evolution idea in the unknown environments has been studied in this paper. The initial population of feasible paths is generated by many isomorphic robots. At the same time some key intersection points has be recorded based on some individuals of each generation population, which may produce some important paths as a beneficial supplement to the current population. Finally a multi-robot cooperative evolution path planning strategy with variable-length genome is presented.Thirdly, a multi-objective path planning strategy based on co-evolution idea in an unknown environment is proposed in this paper. The initial population of feasible paths are built through that multi-robots perform their available path generation method respect respectively and repeated. Based on that, co-evolution includes inside co-evolution of a sub-population and cross evolution operation of inter sub-population. The genes in a sub-population could update information of genes from other subgroup by the improved key-point crossover operator, as a beneficial supplement to the current subgroups.Fourthly, in combination with the first two chapters of the unknown environment based on co-evolutionary strategy research. a multi-objective path planning strategy based on co-evolution idea in community perception networks was constructed, by arranging the static sensor nodes in the working environment. The path planning was designed in dead zone based the behavior were devised through the behavior dynamics method and the path planning was designed in community perception networks were proposed by employing grid method. The initial populations of feasible paths are built through that combination of the two methods.And in the process of evolution, each subgroup of individuals is local co-evolututed optionization in its path section in the perception community?...