The Application Of IGA Based On Similarity And Vector Distance In Soccer Robot Decision-making Subsystems Coordination Layer

In recent years, soccer robot competition has been a high-tech counteractivity which develops rapidly around the world. Soccer robot competitionsystem involves many domains, such as robot theory, image processing,automatic control, correspondence, sensing, and artificial intelligence and so on.At present, it has become an important experimental platform for study onmulti-agent system and artificial intelligence technology.Looked from computation angle, biological immune system is highlyparallel, distributed, auto-adapted and auto-organized, and which has strongability of recognition, study and memory. In recent years, artificial immunealgorithm is emerging which is inspired by biology immune system. As a newresearch domain of intelligent computation, artificial immune algorithm hasalready been used in domains such as machine learning, breakdown diagnosis, image processing, multi-agent decision system, network invasion examination.It has displayed remarkable performance and efficiency. This paper takesMiroSot's middle league simulation system—Middle League SimuroSot as ourexperiment platform, and does some research work which are used artificialimmune algorithm theory and application.The main contents are as follows:(1) The paper analyzes four parts of MiroSot robot soccer competitionsystem: the vision subsystem, the decision-making subsystem, thecommunication subsystem and the robot subsystem. And the paper simplyintroduces their work mechanism. Then the paper studies on the core of MiroSotsoccer robot competition system—decision-making subsystem. This paper usesthree layers control and decision-making model of decision-making subsystemwhich is used widely at present. This model divides the decision-makingsubsystem into coordinated layer, motion planning layer, and elementary actionlayer. Importantly, the paper aims at two major parts of coordination layer (theanalysis about situation and the role assignment). This article unifies theadvantage and disadvantage of the way of dividing the competition field and therelative position method in the situation analysis part, and proposes thecomprehensive factor analytic method. That is, it districts competition locationaccording to certainly regular, and considers real-time condition information ofthe competition simultaneously. At the role assignment part, this paper proposesauto-organized attack and defense strategy which is based on Key-player, and discusses the ball in different regions of the competition field separately, andassigns specific roles for each robot.(2) On the base of analyzing biology immune system's function,mechanism and characters, the paper analyzes Immune Genetic Algorithm (IGA)with more emphasis and introduces IGA based on similarity and vector distanceand proves its convergence. This algorithm unifies local and global searchingthrough selection probability of similarity and vector distance. It can retain mostsuperior antibody, and simultaneously it guarantees the multiplicity of antibody.And premature convergence to this algorithm is overcome. This paper provesthis algorithm's validity and superiority through comparing effects of thefunction optimization experiment and TSP simulation experiment with the effectof Genetic Algorithm (GA).(3) The paper applies IGA based on of similarity and vector distance tocoordination layer of the soccer robot decision-making subsystem, and givesspecific steps to realize it. Through simulation experiment in Middle LeagueSimuroSot simulation platform—Robot Soccer vl.5a, the paper proves thatapplying this algorithm to coordination layer of decision -making subsystem isfeasible and performed well.