Semiotics and Intelligent Control

The overall purpose of this paper is to demonstrate the relevance of semiotics concepts to the analysis of intelligent control systems. Semiotics has only a minor impact on research within intelligent control or robotics. These areas are currently dominated by mathematical concepts of control theory and information processing concepts of artificial intelligence. This situation is unfortunate because the understanding of a complex control problem requires an analysis of the sign relations between sensory data and their meanings and the sign relations between the physical result of a control action and the intentions of control agents. These problems of sign interpretation inherent in all control situations are relevant for the design of automated controls and of the interaction between human operators and the automation. The relevance of semiotics to control problems is demonstrated by applying the semiotics of action developed by Charles Morris to various aspects of a control situation. We use Morris’s definition of three dimensions of signifying that arises from a decomposition of an action into perceptual, appraisive and prescriptive stages. These distinctions identify types of knowledge that a control agent must have in order to cope with a control situation in the environment. Selected examples from the domains of robotics and process control are used to demonstrate that Morris semiotics of action is valuable for conceptual analysis of control situations and for systems design. 1. A SEMIOTIC ANALYSIS OF ACTIONS Morris (1985) semiotic of action describe how an agent interacting with an environment associate meaning to events. It is assumed in the analysis that the agent interacts with the environment in order to reach goals that in some way are valuable or desirable for the agent. The courses of action chosen by the agent depend both on these goals and on observations of the state of the environment. Morris’s semiotics is based on Mead’s analysis of the act (Mead 1938) and makes a distinction between three dimensions of significations of signs namely a designative, an appraisive and a prescriptive dimension. These three dimensions of signification are derived from Mead’s decomposition of an act into three phases. Morris and Mead discuss the behavior of biological species i.e. they talk about the acts of an organism. We will generalize this to agents, which could be biological organisms (e.g. humans) or artifacts (robots). It should be stressed that we do not associate symbolic activity with artifacts. The relations between signs in the environment and their meaning or signification, should be understood as modes of interpretation of events in the environment that are prescribed by the designer and implemented in the hardware or software of the robot. Morris’s analysis is summarized in table 1. 1 Presented at: IFIP WG8.1 Working Conf. Organizational Semiotics. Montreal, Canada, July 23 -25, 2001