An organization for intelligent robot control

We present an organization of an intelligent robot control system. We develop the organization in increments, starting with basic concepts in systems and control. We then include concepts of intelligent, knowledge-based control, followed by concepts of hierarchical control. In the basic systems and control concepts we develop an organization of a controller as an observer and an actor, sharing information about the state of the robot and its world. The actor plans and implements actions that change the world state and the observer plans and implements observations that change an internal representation of the world state. In the intelligent, knowledge-based control concepts we focus on the knowledge that supports the observer and actor, which we call a world model. The world model contains an internal representation of the changing state of the external world (WHAT/WHERE/WHEN knowledge). It also contains representations of the goals and constraints of actions and observations (WHY knowledge). The world model also contains representations of the effect of actions on the world state and the effect of observations on the internal world state model, along with procedures for selecting and implementing goal-satisfying actions and observations (HOW knowledge). In the development of the world model, we emphasize the role of the internal state model for integrating actor and observer operation, and we describe how goal knowledge has a role in coordinating actor and observer operation. In the hierarchical control concepts, we consider hierarchical organizations of actors and observers and study the world model knowledge structures that support them. We emphasize the important role of the world model in integrating the operation of complex, hierarchically organized control systems. The result of our development of an intelligent robot control organization is a series of perspectives that each focus on key aspects of control, while supporting a broad framework for integration of the perspectives into a coherent control system. The organization thus provides the basis of a powerful reference model to aid in the definition, design, and development of complex robot control systems.