Ball Passing: Balancing Rewards, Risks, Costs, and Real-Time Constraints

We are looking a generic solution for the optimized ball passing problem in the robotic soccer which is applicable to different RoboCup leagues and other digital simulated sports games like basketball or ice hockey. In doing so, we show that previously published ball passing methods do not properly address the necessary balance between the anticipated rewards, costs, and risks. The multi-criteria nature of this optimization problem requires using the Pareto optimality approach. The problem itself is substantially inconvex, nothing else except the search of all available alternatives in the Pareto set appears to be applicable in this case. Real-time constraints are further complicating the problem. We propose a scalable and robust solution for decision making with multiple optimality criteria; is quality degrades in a graceful way once the real time constrains are kicking in. Our method is treating equally direct and leading passes to the partners and self passing while fast dribbling the ball. The new method also allows easily modeling the whole spectrum of risk aversive to risk talking attitudes; therefore it is generic indeed.