Autor(es):
Ribeiro, António Fernando Alcântara
Data: 2024
Identificador Persistente: https://hdl.handle.net/1822/93742
Origem: RepositóriUM - Universidade do Minho
Assunto(s): Football Strategy; Multi-Agent, Autonomous Robot; RoboCup; Probability-Based; Play Generator; Decision Trees; Heat Maps; Simulation; Real-Time Communication; Estratégia de Futebol; Multi-Agente; Robô Autónomo; Baseado em Probabilidade; Gerador de Jogadas; Árvores de Decisão; Mapas de Probabilidades; Simulação; Comunicações em Tempo Real
Descrição
RoboCup is the Robotics World Championship and is one of the reasons for the constant growth, research and development in robotics. One of its leagues is the Middle Size League (MSL), which is a scientific football challenge with teams of five robots. The strategy of a multi-autonomous cooperative robotic system in a football game can be solved in multiple ways, commonly based on the Skills, Tactics and Plays (STP) architecture. With a different approach regarding its architecture, a centralised Probability- Based Strategy (PBS) method is presented in this dissertation and implemented in the LAR@MSL team from the University of Minho. It uses autonomous skills, generates tactics and plays in real-time, based on the probability values of every action possible. The action probabilities also position each robot to optimise the overall probability. This new method centralises all the decision-making processes in the base station, to which the robots communicate, transmit all the data acquired, and receive the ideal skill and respective arguments. The robots are still fully autonomous regarding the skills given to them, just like the strategy improvements and calibration are independent of the robots’ hardware and control systems. Also crucial for this solution was the development of the simulator environment, the communication system and the base station interface. The simulator greatly improved the development time for the strategy, allowing the strategy to be developed in parallel with the robot’s hardware. The communications system implemented was tested and proved reliable even in the most saturated environments. The base station interface developed is the primary debugging tool that allowed the team to interpret values in real-time and interact and calibrate both robots and the strategy. The strategy presented has been fully implemented within the team and tested in multiple scenarios, such as simulators, a controlled environment, against humans in a simulator, and at the RoboCup competition. The Probability-Based Strategy (PBS) proved modular, flexible, and easily adaptable to different game situations with minimal effort. It allows different behaviours against different opponents and can support any team size, either its own or the opponent’s team, without the need for a predefined set of plays.
