Real-time intelligent behaviour in dynamic environments : soccer-playing robots

An autonomous robot operating in a dynamic environment is confronted by the question of, "What to do now?" The problem of designing a controller for a car-like robot competing with another robot in a game of soccer is considered. This is a dynamic environment; the locations of the ball and the robots are constantly changing. Rapid and appropriate responses to changes in the world are central to intelligent behaviour. There are no established robot architectures that seem adequate for the challenges posed in dynamic domains. Traditional work in Artificial Intelligence has focused on the construction of plans for future execution and has resulted in architectures that are not extensible to dynamic environments. Even recently developed "reactive" architectures such as the subsumption architecture, situated automata, and RAP do not seem satisfactory. Reactive deliberation has been designed to present a set of structural elements needed in dynamic domains. Reactive deliberation makes three contributions to robot architecture. First, the decisions of what task to achieve and how to achieve it are best resolved in unison. Second, the transient goals of a robot must be evaluated at a rate commensurate with changes in the environment. Third, goal-oriented modules called behaviours are a useful abstraction that allow effective goal-arbitration and sharing of scarce computational resources. The effectiveness of reactive deliberation has been demonstrated through a tournament of one-on-one soccer games between robots. Current functionality includes motion planning, ball shooting, and playing goal, with accurate motion control at speeds of 1 m/s. The results of the soccer tournament suggest that the architectural elements in reactive deliberation are sufficient for real-time intelligent control in dynamic environments.