Strategies in Robot Soccer Game

Abstract

Robot soccer is a platform of multidisciplinary integration, especially it integrates artificial intelligence, computer and robotics together. It promotes the development of autonomous multi-agent robotic system, contributes to the state-of-the-art technology improvement in robotics and artificial intelligence, helps intercrossing between disciplines. Robot soccer studies how mobile robots can be built and trained to play a game of soccer. At present, there are two most important international robot soccer competitions, named FIRA Cup and RoboCup. Soccer is a team sport where the collaboration is very important. Robot soccer requires not only cooperation between robots, but also collaboration with high efficiency. In robot soccer, robots stay in a dynamic environment all the time. The teammate robots need to gradually gain ascendency over opponents and finally win the game through making right decisions and taking appropriate actions according to the changing situation. In this process, strategy plays an important role. Our work contributes to dealing with the construction, evaluation, and optimization of strategy and strategy set in robot soccer. We divide the robot soccer game into strategy level and tactic level. On the strategy level, robots make plans to organize teamwork. On the tactic level, robots take actions to implement the strategies. A strategy is a plan consists of current condition and corresponding response actions. A number of strategies constitute strategy set. Strategy set is the kernel of teamwork in robot soccer game. Tactics could be conceived as the implementation of strategies. The tactic covers task allocation, path planning, obstacle avoidance, collision detection, collision prediction, trajectory prediction, motion control, and so on.