Flexible positioning of secondary attackers based on adaptive neuro-fuzzy inference system

RoboCup is a standard problem of multi-agent systems (MAS) so that various theories, algorithms and architectures can be evaluated. A flexible positioning of secondary attackers based on adaptive neuro-fuzzy inference system (ANFIS) is presented. It is a new approach to the positioning problem in RoboCup Simulation. First, the network architecture and learning algorithms of ANFIS are briefly introduced. Then, based on ANFIS, the flexible positioning of secondary attackers is presented to train the agents. Finally, a hybrid learning rule which combines the gradient method and the least squares estimate (LSE) to identify parameters is proposed. It cuts down the convergence time of ANFIS substantially.