The Dynamic Object Path Planning Based On Penalty Function Niche Pareto Genetic Algorithm

Presently, along with rapid development of computer network technology, the computer fast changes to the opening, network platform, cooperation working mode. Agent based theory and technical, especially multi-agent systems, bring us a brand-new mode to design and implement software system which run in a distribution and opening environment.RoboCup (The RobotWorld Cup Initiative) is a typical multi-agent system, has currently chosen soccer as its standard task. In the RoboCup, the purpose of the path planning is to find a path without collision which satisfied some estimate criterions. The path planning is applied in the lower policy which uses as the foundation of the robots' basic actions, the planning result directly has an effect on the action's real-time and the veracity. So, everyone who study on the robot look path planning as an emphases.On the base of analyzing the traditional path planning methods of the robots, the paper looks path planning as a multi-object optimize problem. We summarize the stamina model of the robot as an ternary set, and then we set the athletics model which can planning the path for the dynamic object. At last we give the planning algorithm based on the penalty function niche pareto genetic algorithm. The innovation and the main work of the paper list as bellows:(1)On the base of detailedly analyzing the SoccerServer of the robot's stamina, we bring forward the robot's stamina model: a ternary set. The using efficiency of the player, the stamina resume rate of the player and the needed time of path planning. And we regard stamina model as an object of the multi-object optimize.(2)In this paper we bring forward an athletics model which aim at dynamic object path planning, and we combine with the stamina model. At last we use the niche pareto genetic algorithm based on penalty function.(3)On the base detailedly analyzing genetic algorithm and niche pareto genetic algorithm base theoretics, then analyzing the three standard realizations of the niche pareto genetic algorithm(NPGA). In the paper we bring forward a penalty function, so we can assure that in a niche there will be only one best root. So we can avoid constringency local classic root, and accelerate the constringency speed.