Action 35 and Planning in Embedded Agents *

Embedded agents are computer systems that sense and act on their environments, monitoring complex dynamic conditions and affecting the environment in goal-directed ways. Systems of this kind are extremely difficult to design and build, and without clear conceptual models and powerful programming tools, the complexities of the real world can quickly become overwhelming. In certain special cases, designs can be based on wellunderstood mathematical paradigms such as classical control theory. More typically, however, tractable models of this type are not available and alternative approaches must be used. One such alternative is the situated-automata framework, which models the relationship between embedded control systems and the external world in qualititative terms and provides a family of programming abstractions to aid the designer. This paper briefly reviews the situated-automata approach and then explores in greater detail one aspect of the approach, namely the design of the action-generating component of embedded agents.