Abstract not found

This paper describes the main properties of behavior-based approaches to control. Different approaches to designing and using behaviors as basic units for control, representation, and learning are illustrated on three empirical examples of robots performing navigation and path-finding, group behaviors, and learning behavior selection. 1 Introduction An architecture provides a set of principles for organizing control systems. In addition to supplying structure, it imposes constraints on the way control problems can be solved. In this paper we explore the constraints of behavior-based approaches to control, and demonstrate them on three architectures that were used to implement robots that successfully performed navigation and pathfinding, group behaviors, and learning of behavior selection. In each case, we focus on the different ways behaviors are defined, modularized, and combined. This paper is organized as follows. Section 2 gives an overview of basic approaches to autonomous agent...

We present data and argument to show that in Tetris-a real-time, interactive video game-certain cognitive and perceptual problems ore more quicktv, easily, and reliably solved by performing actions in the world than by performing com-putational actions in the head atone. We have found that some of the translations and rotations made by players of this video game are best understood as actions that use the world to improve cognition. These actions are not used to implement a plan, or to implement a reaction; they are used to change the world in order to simplify the problem-solving task. Thus, we distinguish pragmatic octions--actions performed to bring one physically closer to a goal-from epistemic actions-actions performed to uncover informatioan that is hidden or hard to compute mentally. To illustrate the need for epistemic actions, we first develop a standard information-processing model of Tetris cognition and show that it cannot explain performance data from human players of the game-even when we relax the assumption of fully sequential processing. Standard models disregard many actions taken by players because they appear unmotivated or superfluous. How-ever, we show that such actions are actually far from superfluous; they play a valuable role in improving human performance. We argue that traditional accounts are limited because they regard action as having o single function: to change the world. By recognizing a second function of action-an epistemic func-tion-we can explain many of the actions that a traditional model cannot. Al-though our argument is supported by numerous examples specifically from Tetris, we outline how the new category of epistemic action can be incorporated into theories of action more generally. In this article, we introduce the general idea of an epistemic action and discuss its role in Tetris, a real-time, interactive video game. Epistemic actions-physical actions that make mental computation easier, faster, or more We thank Steve Haehnichen for his work on the initial implementations of Tetris and

To describe phenomena that occur at different time scales, computational models of the brain must necessarily incorporate different levels of abstraction. We argue that at time scales of approximately one-third of a second, orienting movements of the body play a crucial role in cognition and form a useful computational level, termed the embodiment level . At this level, the constraints of the body determine the nature of cognitive operations, since the natural sequentiality of body movements can be matched to the natural computational economies of sequential decision systems. The way this is done is through a system of implicit reference termed deictic, whereby pointing movements are used to bind objects in the world to cognitive programs. We show how deictic bindings enable the solution of natural tasks and argue that one of the central features of cognition, working memory, can be related to moment-by-moment dispositions of body features such as eye movements and hand movements. Keyw...

The congeries of theoretical views collectively referred to as "situated action" (SA) claim that humans and their interactions with the world cannot be understood using symbol-system models and methodology, but only by observing them within real-world contexts or building nonsymbolic models of them. SA claims also that rapid, real-time interaction with a dynamically changing environment is not amenable to symbolic interpretation of the sort espoused by the cognitive science of recent decades. Planning and representation, central to symbolic theories, are claimed to be irrelevant in everyday human activity. We will contest these claims, as well as their proponents ' characterizations of the symbol-system viewpoint. We will show that a number of existing symbolic systems perform well in temporally demanding tasks embedded in complex environments, whereas the systems usually regarded as exemplifying SA are thoroughly symbolic (and representational), and, to the extent that they are limited in these respects, have doubtful prospects for extension to complex tasks. As our title suggests, we propose that the goals set forth by the proponents of SA can be attained only within the framework of symbolic systems. The main body of empirical evidence supporting our view resides in the numerous symbol systems constructed in the past 35 years that have successfully simulated broad areas of human cognition. During the past few years a point of view has emerged in artificial intelligence, often under the label of "situated action " (henceforth, SA), that denies that intelligent systems are correctly characterized as physical symbol systems, and especially denies that symbolic processing lies at the heart of

Abstract not found

This paper presents U-Tree, a reinforcement learning algorithm that uses selective attention and shortterm memory to simultaneously address the intertwined problems of large perceptual state spaces and hidden state. By combining the advantages of work in instance-based (or "memory-based") learning and work with robust statistical tests for separating noise from task structure, the method learns quickly, creates only task-relevant state distinctions, and handles noise well. U-Tree uses a tree-structured representation, and is related to work on Prediction Suffix Trees [Ron et al., 1994] , Parti-game [Moore, 1993] , G-algorithm [Chapman and Kaelbling, 1991] , and Variable Resolution Dynamic Programming [Moore, 1991] . It builds on Utile Suffix Memory [McCallum, 1995c] , which only used short-term memory, not selective perception. The algorithm is demonstrated solving a highway driving task in which the agent weaves around slower and faster traffic. The agent uses active perception with ...

All intelligence relies on search --- for example, the search for an intelligent agent's next action. Search is only likely to succeed in resource-bounded agents if they have already been biased towards finding the right answer. In artificial agents, the primary source of bias is engineering. This dissertation

This paper reports on experiments where techniques of supervised machine learning are applied to the problem of planning. The input to the learning algorithm is composed of a description of a planning domain, planning problems in this domain, and solutions for them. The output is an efficient algorithm --- a strategy --- for solving problems in that domain. We test the strategy on an independent set of planning problems from the same domain, so that success is measured by its ability to solve complete problems. A system, L2Act, has been developed in order to perform these experiments. We have experimented with the blocks world domain, and the logistics domain, using strategies in the form of a generalization of decision lists, where the rules on the list are existentially quantified first order expressions. The learning algorithm is a variant of Rivest`s [39] algorithm, improved with several techniques that reduce its time complexity. As the experiments demonstrate, generalization is a...

We formalize a model for supervised learning of action strategies in dynamic stochastic domains and show that PAC-learning results on Occam algorithms hold in this model as well. We then identify a class of rule-based action strategies for which polynomial time learning is possible. The representation of strategies is a generalization of decision lists; strategies include rules with existentially quantified conditions, simple recursive predicates, and small internal state, but are syntactically restricted. We also study the learnability of hierarchically composed strategies where a subroutine already acquired can be used as a basic action in a higher level strategy. We prove some positive results in this setting, but also show that in some cases the hierarchical learning problem is computationally hard. 1 Introduction We formalize a model for supervised learning of action strategies in dynamic stochastic domains, and study the learnability of strategies represented by rule-based syste...