Muitos problemas de otimização em ambientes multiagentes utilizam os algoritmos evolutivos para encontrar as melhores soluções. Uma das abordagens mais utilizadas consiste na aplicação de um algoritmo genético, como alternativa aos métodos tradicionais, para definir as ações dos jogadores em um time de futebol de robôs. Entretanto, conforme relatado na literatura, há inúmeras possibilidades e formas de se aplicar um algoritmo genético no domínio do futebol de robôs. Assim sendo, neste trabalho buscou-se realizar uma análise comparativa dos algoritmos genéticos mono-objetivo e multi-objetivo aplicados no domínio do futebol de robôs. O problema padrão escolhido para realizar essa análise foi de desenvolver uma estratégia de controle autônomo, a fim de capacitar que os robôs tomem decisões sem interferência externa, pois, além de sua solução se encontrar ainda em aberto, o mesmo é também de suma relevância para a área de robótica. / Many optimization problems in multiagent environments adapt evolutionary algorithms to find the best solutions. A widely used approach consists of applying a genetic algorithm as an alternative to traditional methods, in order to define the actions of the players on a soccer team of simulated robots. However, as reported in the literature, there are many possibilities and ways to apply a genetic algorithm in the field of robot soccer. Therefore, this work attempts to make a comparative analysis of mono-objective and multi-objective genetic algorithms applied to control a robot soccer. The standard problem chosen for this analysis was to develop a strategy for autonomous control, in order to enable the robots to make decisions without external interference, because in addition to its solution is still open, it is also of utmost relevance to the area robotics.

Muitos problemas de otimização em ambientes multiagentes utilizam os algoritmos evolutivos para encontrar as melhores soluções. Uma das abordagens mais utilizadas consiste na aplicação de um algoritmo genético, como alternativa aos métodos tradicionais, para definir as ações dos jogadores em um time de futebol de robôs. Entretanto, conforme relatado na literatura, há inúmeras possibilidades e formas de se aplicar um algoritmo genético no domínio do futebol de robôs. Assim sendo, neste trabalho buscou-se realizar uma análise comparativa dos algoritmos genéticos mono-objetivo e multi-objetivo aplicados no domínio do futebol de robôs. O problema padrão escolhido para realizar essa análise foi de desenvolver uma estratégia de controle autônomo, a fim de capacitar que os robôs tomem decisões sem interferência externa, pois, além de sua solução se encontrar ainda em aberto, o mesmo é também de suma relevância para a área de robótica. / Many optimization problems in multiagent environments adapt evolutionary algorithms to find the best solutions. A widely used approach consists of applying a genetic algorithm as an alternative to traditional methods, in order to define the actions of the players on a soccer team of simulated robots. However, as reported in the literature, there are many possibilities and ways to apply a genetic algorithm in the field of robot soccer. Therefore, this work attempts to make a comparative analysis of mono-objective and multi-objective genetic algorithms applied to control a robot soccer. The standard problem chosen for this analysis was to develop a strategy for autonomous control, in order to enable the robots to make decisions without external interference, because in addition to its solution is still open, it is also of utmost relevance to the area robotics.

Agent-based systems usually operate in real-time, stochastic and dynamic environments. Many theoretical and applied techniques have been applied to the investigation of agent architecture with respect to communication, cooperation, and learning, in order to provide a framework for implementing artificial intelligence and computing techniques. Intelligent agents are required to be able to adapt and learn in uncertain environments via communication and collaboration (in both competitive and cooperative situations). The ability of reasoning and learning is one fundamental feature for intelligent agents. Due to the inherent complexity, however, it is difficult to verify the properties of the complex and dynamic environments a priori. Since analytic techniques are inadequate for solving these problems, reinforcement learning (RL) has appeared as a popular approach by mapping states to actions, so as to maximise the long-term rewards. Computer simulation is needed to replicate an experiment for testing and verifying the efficiency of simulation-based optimisation techniques. In doing so, a simulation testbed called robot soccer is used to test the learning algorithms in the specified scenarios. This research involves the investigation of simulation-based optimisation techniques in agent-based systems. Firstly, a hybrid agent teaming framework is presented for investigating agent team architecture, learning abilities, and other specific behaviors. Secondly, the novel reinforcement learning algorithms to verify goal-oriented agents; competitive and cooperative learning abilities for decision-making are developed. In addition, the function approximation technique known as tile coding (TC), is used to avoid the state space growing exponentially with the curse of dimensionality. Thirdly, the underlying mechanism of eligibility traces is analysed in terms of on-policy algorithm and off-policy algorithm, accumulating traces and replacing traces. Fourthly, the "design of experiment" techniques, such as Simulated Annealing method and Response Surface methodology, are integrated with reinforcement learning techniques to enhance the performance. Fifthly, a methodology is proposed to find the optimal parameter values to improve convergence and efficiency of the learning algorithms. Finally, this thesis provides a serious full-fledged numerical analysis on the efficiency of various RL techniques.

This work describes a design of an agent model for robotic soccer based on the DEVS formalism. There is also presented a design of own DEVS simulator (based on classic DEVS simulator) for parallel realtime simulations. Functionality of the simulator and the model is shown on an example of a soccer client for RoboCup Soccer Server. Based on this client, there is also presented a design of a library for easier creation of soccer clients for RoboCup.