Reinforcement learning on strategy selection for a cooperative robot system

The article presents a multi-strategy decision making system for robot soccer games. Through reinforcement processes, the coordination and cooperation between robots are learned on the flight of game. Meanwhile, a better action can be granted after iterative learning process. The experimental scenario is a 5-on-5-soccer game where the proposed system dynamically assigns each player a primitive role, such as attacker, goalie, etc. The responsibility of each role varies along state transitions. Therefore, the system consists of several strategies, such as offensive strategy, defensive strategy, and so on. As well, the decision-making mechanism can choose a better strategy in turns under circumstances encountered. In each strategy, a robot should behavior cooperatively with teammates and resolve conflicts aggressively. The major task assignment to robots in each strategy is simply to catch good positions. Therefore, the problem of dispatching robots to good positions in a reasonable manner should be coped with. This kind of problems is similar to dispatch problems in linear programming research. Utilizing modified simplex and branch and bound methods, each robot can be granted to its assigned spot with minimal cost. Consequently, robots based on the proposed decision making system can accomplish each situational task cooperatively