Multi-Robot Path-Planning Based On Genetic Algorithms

The development of multi-robot system is based on communication, control theory, computer technology, image process, artificial intelligence, etc. Soccer robot is a typical multi-agent system, which system is dynamic and collaborative between multi-agents. We can research and evaluate various theories and methods deeply in the mutli-agent system.Microsoft simulation robot soccer system is used as the patlform in this paper. The system structure of Microsoft simulation robot soccer is analyzed and the decision system is studied. This paper proposes a four- layer decision-making model based on partition according to the decision system, wchich is composed of information processing layer, district decision-making layer, path-planning layer and motion layer, and indicates that there are some problems in the implement of path-planning layer. If the shortest path is the only principle in the implement, the robots are easy to contradict each other, collided the other robot and fly away in the competition field and this will clog the preceeding of the soccer and affect the result of the soccer. This paper proposes a multi-robot path-planning based on genetic algorithms to resolve the problem.This paper proposes a path optimization strategy based on genetic algorithms. The optimization target of the strategy is to make the path is obstacle-avoidance and shortest. The target is used to design the fitness function based on genetic algorithms. In order to optimize the path, initial path population is generated and select, cross, muate the path in the population, then add smooth, delete and insert operator in consideration of continuity of the path which is composed of ployline, and use principle of natural selection, which is that survive in the fittest,to seclet the optimatal path which is the largest fitness value in the last generation.This paper applys the path optimization strategy based on genetic algorithms in Microsoft simulation robot soccer platform. Firstly, the strategy is used in the static path planning of soccer robot. Based on the applyment, the new algorithm which is united of genetic algorithms and the shortest algorithms is to resolve the dynamic path planning of multi-robots in static method. When the robot needs to achieve the new target point, the two methods are to be judged which is the better algorithm to plan the path in the new environment and the robot finally achieves the target point quickly and effectively. The goal to make the path obstacle-avoidance and shortest is implemented in the path-planning layer.This paper applys the static path-palnning and the dynamic path-planning based on genetic algorithms in the simulation platform and analyzes the experiment result. The algorithm is verified to be obstacle-avoidance and shortest compared to other algorithms...