Research On Task Allocation For Robot Soccer System

Driven by the needs of modern science and technology development and the national economy, robot technology is playing an increasingly important role in real life; at the same time, requirements on the robot are increasing highly, and a lot of work can not just rely on a single robot,therefore, the people carrying out the research and application of multi-robot system. Compared with a single robot, multi-robot systems have been greatly improved in the using scope, quality of the done job, technology requirements and expansion of space,if all the robots in the system can collaborate harmoniously.A lot of research on the task allocation of multi-robot soccer system is done,researchers either try to find the new algorithm for task allocation,or improve the existing algorithms continually, and establish many international organizations and game platform to promote its ongoingresearch and development.Basing on the research status at home and abroad,this paper carries out the following main contents:1.Genetic algorithm improvement for multi-robot soccer game systemAccording to the disadvantage of prematurity and slow convergence of the traditional algorithm, many improvements of the algorithm in all respects have been made,and MATLAB simulation proves the great efficiency of the improved algorithm, getting ready for the application in multi-robot soccer system. Genetic algorithm can be implemented in the multi-robot soccer system easily, and the superiority of the algorithm will be more obvious with the increase in the number of robots and tasks.2. Implementation of improved genetic algorithm for task allocation in multi-robot soccer systemProposing a new method of initializing population,in which individual genes are determined by the utility value,and good genes can be re-used, to ensure that good genes can be retained even after repeated evolution and avoid the blindness of generating initial population randomly; Hierarchical genetic algorithm is used for groups with different optimal genes to evolute separately,till a certain algebra, elect a certain number of the best individuals to go on evoluting, in order to achieve the combination of multi-optimal gene. Simulation results demonstrate the superiority of the improved algorithm.Finally,based on the research above, the paper sums up,and pointes out the deficiencies and areas for further research.