Robot Soccer Strategy Based on Hierarchical Finite State Machine to Centralized Architectures

Abstract

[EN] Coordination among the robots allows a robot soccer team to perform better through coordinated behaviors. This requires that team strategy is designed in line with the conditions of the game. This paper presents the architecture for robot soccer team coordination, involving the dynamic assignment of roles among the players. This strategy is divided into tactics, which are selected by a Hierarchical State Machine. Once a tactic has been selected, it is assigned roles to players, depending on the game conditions. Each role performs defined behaviors selected by the Hierarchical State Machine. To carry out the behaviors, robots are controlled by the lowest level of the Hierarchical State Machine. The architecture proposed is designed for robot soccer teams with a central decision-making body, with global perception. 200 games were performed against a team with constant roles, winning the 92.5% of the games, scoring more goals on average that the opponent, and showing a higher percent of ball possession. Student s t-test shows better matching with measurement uncertainty of the strategy proposed. This architecture allowed an intuitive design of the robot soccer strategy, facilitating the design of the rules for role selection and behaviors performed by the players, depending on the game conditions. Collaborative behaviors and uniformity within the players behaviors during the tactics and behaviors transitions were observed