Research On Path Planning For Soccer Robot Based On Improved Genetic Algorithm

Soccer robot system is a multi-agent system, It’s very representative and typical. The environment of soccer robot system is rather complicated, It needs to meet real-time dynamic demand as well as to fight against the opposite robots. In artificial intelligence, it is a good experimental platform on the theoretical study and model test. But path planning of soccer robot must need to realize effectively obstacle avoidance, and use the shortest time to reach the destination. Thus it can be seen that:path planning of soccer robot as a research topic is very imporant, but also very challenging.There are a lot of research methods on path planning, such as grid method, artificial potential field,visibility graph and some methods of artificial intelligence, here intelligent algorithm has neural network method, genetic altorithm,etc. Each of them has its advantages and disadvantages, but it is not very perfect to realize a good path planning. So we need further improvement to a certain extent.In this paper, experimental platform of robot soccer system is MiroSot3:3. Here research method of path planning for soccer robot is to improve GA based on the basic genetic algorithm, then in-depth study. Thus we can explore a feasible path planning method of soccer robot, and prove its validity in the practical application.Firstly, we introduce the development history of soccer robot, research significance of soccer robot and its path planning, the typical path planning methods and its research status at home and abroad.Secondly, soccer robot system and each subsystem system are introduced in brief, and environmental model is established, including pitch model and robot model.Then, genetic algorithm is introduced in detail, including the basic theory, the main components and the basic steps, we also introduce the application of genetic algorithm to path planning for soccer robot. When solving the process in the traditional genetic algorithm, it is easy to fall into local optimum, and slow convergence speed and other shortcomings. So in this paper we improve it based on GA, here we come up with adaptive genetic algorithm which combining adaptive algorithm with GA. Aiming at several kinds of existing adaptive genetic algorithm, due to the design of genetic operator, the evolutionary of individual that fitness value is very large has disadvantage. So on the basis of it we improve it, thus it can avoid falling into local optimal solution, enhance the ability of global optimization, and improve the rate of convergence.Finally, the improved adaptive genetic algorithm is applied to path planning for soccer robot, we use the Matlab tool to do simulation experiment, and see whether experimental curves accord with requirement. Next we use computer simulation platform to play game in simulation, competition venues and robot shape can be vividly imitate, and the game screen we see is very clear, we can verify that the path planning method is effective. At last, it has been applied in actual practice.