Integrating Reactivity, Goals, and Emotion in a Broad Agent

Researchers studying autonomous agents are increasingly examining the problem of integrating multiple capabilities into single agents. The Oz project is developing technology for dramatic, interactive, simulated worlds. One requirement of such worlds is the presence of broad, though perhaps shallow, agents. To support our needs, we are developing an agent architecture, called Tok, that displays reactivity, goal-directed behavior, and emotion, along with other capabilities. Integrating the components of Tok into a coherent whole raises issues of how the parts interact, and seems to place constraints on the nature of each component. Here we describe brie y the integration issues we have encountered in building a particular Tok agent (Lyotard the cat), note their impact on the architecture, and suggest that modeling emotion, in particular, may constrain the design of integrated agent architectures. 1 BROAD AGENTS 1 1 Broad Agents The Oz project [3] at Carnegie Mellon is developing technology for artistically interesting, highly interactive, simulated worlds. We want to give users the experience of living in (not merely watching) dramatically rich worlds that include moderately competent, emotional agents. An Oz world has four primary components. There is a simulated physical environment, a set of automated agents which help populate the world, a user interface to allow one or more people to participate in the world [13], and a two-player adversary search planner concerned with the long term structure of the user's experience [2]. Oz shares some goals with traditional story generation systems [18, 16], but adds the signi cant requirement of rich interactivity. One of the keys to an artistically engaging experience is for the user to be able to \suspend disbelief". That is, the user must be able to imagine that the world portrayed is real, without being jarred out of this belief by the world's behavior. The automated agents, in particular, mustn't be blatantly unreal. Thus, part of our e ort is aimed at producing agents with a broad set of capabilities, including goal-directed reactive behavior, emotional state and behavior, and some natural language abilities. For our purpose, each of these capacities may be as shallow as necessary to allow us to build broad, integrated agents [4]. Oz worlds are far simpler than the real world, but they must retain su cient complexity to serve as interesting artistic vehicles. The complexity level is somewhat higher, but not exceptionally higher, than typical AI micro-worlds. Despite these simpli cations, we nd that our agents must deal with imprecise and erroneous perceptions, with the need to respond rapidly, and with a general inability to fully model the agent-rich world they inhabit. We suspect that some of our experience with broad agents in Oz may transfer to other domains, such as social, real-world robots. Building broad agents is a little studied area. Much work has been done on building reactive systems [6, 10, 9, 24], natural language systems, and even emotion systems [8, 23, 21]. There is growing interest in integrating action and learning (see [14]) and some very interesting work on broader integration [25, 22]. However, we are aware of no other e orts to integrate the particularly wide range of capabilities needed in the Oz domain. Here we present our e orts, focusing on integration mechanisms and their impact on components of the architecture. 2 Tok and Lyotard In analyzing our task domain, we concluded that the capabilities needed in our initial agents are perception, reactivity, goal-directed behavior, emotion, social behavior, natural language analysis, and natural language generation. Our agent architecture,