Multi-Robot Coordination in Highly Dynamic Environments

In recent years, researchers have invested significant effort in multi-robot systems. Robot soccer, as a pioneering multi-robot task, has offered a challenging research testbed. In robot soccer, a team of robots faces an uncertain and dynamic environment including an opponent team.

We have researched in robot soccer, developing single-robot and multi-robot perception, cognition, and action algorithms. To form an effective team, individual robots need to be skilled. We have developed novel object-recognition, localization, and behavior-based algorithms to provide these skills. To achieve a reliable team of robots, we have contributed team coordination strategies, methods for representing and adapting team plans in response to a dynamic world, and multiagent learning. In this talk, I will present our contributions to this complex adversarial multi-robot task, focused in particular on our teams of small wheeled robots and of four-legged Sony AIBO robots.

Based on our research results, I will introduce an overarching hypothesis on a set of principles for effective robot autonomy in the physical world. I will conclude by presenting how my research goals in fit into the broad fields of artificial intelligence and robotics.