Abstract not found

A routine is a habitually repeated performance of some actions. Agents use routines to guide their everyday activities and to enrich their abstract concepts about acts. This dissertation addresses the question of how an agent who is engaged in ordinary, routine activities changes its behavior over time, how the agent's internal representations about the world is affected by its interactions, and what is a good agent architecture for learning routine interactions with the world. In it, I develop a theory that proposes several key processes: (1) automaticity, (2) habituation and skill refinement, (3) abstraction-bychunking, and (4) discovery of new knowledge chunks. The process of automaticity caches the agent's knowledge about actions into a flat stimulus-response data structure that eliminates knowledge of action consequences. The stimulus-response data structure produces a response to environmental stimuli in constant time. The process of habituation and skill refinement uses environm...

In the elephant paper [Bro90], Brooks criticized the ungroundedness of traditional symbol systems, and proposed physically grounded systems as an alternative, in particular the subsumption architecture. Although we are still struggling with many of the issues involved, we believe we have some contributions to make towards solving some of the open problems with physically grounded systems, particularly with respect to whether or how to integrate the old with the new. In this paper we describe an agent architecture that specifies an integration of explicit representation and reasoning mechanisms, embodied semantics through grounding symbols in perception and action, and implicit representations of specialpurpose mechanisms of sensory processing, perception, and motor control. We then present components that we place in our general architecture to build agents that exhibit situated activity and learning, and finally a physical agent implementation and two simulation studies. The gist of o...

If behavior networks, which use spreading activation to select actions, are analogous to connectionist methods of pattern recognition, then recurrent behavior networks, which use energy minimization, are analogous to Hopfield networks. Hopfield networks memorize patterns by making them attractors. Similarly, each behavior of a recurrent behavior network should be an attractor of the network, to inhibit fruitless, repeated switching between different behaviors in response to small changes in the environment and in motivations. I overcome two major objections to this view, and demonstrate that the performance in a test domain of the Do the Right Thing recurrent behavior network is improved by redesigning it to create desirable attractors and basins of attraction. I further show that this performance increase is correlated with an increase in persistence and a decrease in undesirable behavior-switching. On a more general level, this work encourages the study of action selection as a dynam...

Behavior based AI [Brooks, 1990, Maes, 1990] has questioned the need for modeling intelligent agency using generalized cognitive modules for perception and behavior generation. Behavior based AI has demonstrated successful interactions in unpredictable environments in the mobile robot domain [Brooks, 1985, Brooks, 1990]. This has created a gulf between "traditional" approaches to modeling intelligent agency and behavior based approaches. We present an architecture for intelligent autonomous agents which we call GLAIR (Grounded Layered Architecture with Integrated Reasoning) [Hexmoor et al., 1992, Hexmoor et al., 1993b, Hexmoor et al., 1993a]. GLAIR is a general multi-level architecture for autonomous cognitive agents with integrated sensory and motor capabilities. GLAIR offers an "unconscious" layer for modeling tasks that exhibit a close affinity between sensing and acting, i.e., behavior based AI modules, and a "conscious" layer for modeling tasks that exhibit delays between sensing ...

In contrast to "conscious", reasoned behaviors, we consider behaviors that are automatic and unreasoned to be "unconscious". The latter are commonly

Routines are patterns of interaction between an agent and its world. Getting in or out of a car, changing lane, and flipping pages of a book can be routines for an agent if the agent consistently engages in these activities in a similar way. I.e., a task for an agent is a routine if the agent that has choices about how to accomplish that task, nevertheless does it in the same way. Consistently putting on the left leg of pants before putting on the right leg would be a routine for an agent. A routine is either imposed upon the agent (a plan at the conscious level to be followed), in which case it need not be discovered, or performed by the agent automatically, i.e., unconsciously. The latter may or may not ever be discovered, i.e., noticed and made conscious. However, the existence of such a routine may guide the agents actions. If it remains unconscious, it aids in choosing among competing actions unconsciously as an unexplained tendency or a preference. If it is noticed and made consc...