Probability Fuzzy Cognitive Map for Decision-Making in Soccer Robotics

Abstract

Based on research work in soccer robot decision-making, an improved probability fuzzy cognitive map (PFCM) based soccer robot reasoning model (PFCMSRRM) is proposed to overcome limitations of the popular finite-state machine (FSM) approach in decision-making research. Explanation is given on the structure of PFCMSRRM and its decision reasoning implementation. Learning on the weight value of PFCMSRRM is conducted using learning algorithms based on gradient descent and simulated annealing respectively. As a result, we built a PFCMSRRM system used in multi-agents ROBOCUP middle-sized soccer robot competition, with higher accuracy and speed in decision-making as compared with FSM approach, which was proved by experiments.