Real - Time Planning and Intelligent Control

Classical control systems utilize implicit analytical models and select inputs using numerical computation in order to achieve desired plant performance in real time. Artificial intelligence planning systems utilize explicit symbolic or logical models of discrete operations and construct consistent planned sequences using symbolic reasoning to achieve specified task goals. In the development of more complex intelligent systems, these two approaches must interact in order to successfully execute real-time operations in a manner consistent with the global goals and constraints of the task. Intelligent control and real-time planning are approaches to developing a theory and methodology for the design systems which incorporate this interface. In this paper, we provide some examples of the role of task representation in the definition of these control and planning strategies. We describe in detail some results on task planning of assembly sequences and the use of precedence relations as an implicit model of ordering constraints for assembly operations.