L.P. Kaelbling. An architecture for intelligent reactive systems. In Workshop on Planning & Reasoning about Action, pages 235-250, Timberline OR, 1986.  Home/Search   Document Not in Database   Summary   Related Articles   Check

....In recent years, the classical approach to planning has been challenged by many authors (e.g. 1,2,9,12,18,19] The question is whether its fundamental assumptions and complexity make it of any use in real world applications. The arguments against classical planning (e.g. reactive planning [8,26]) are valid in a number of domains. Nevertheless, in our opinion, there are applications where classical planning is still the only reasonable approach, e.g. logistics, process planning, scheduling. Furthermore, in the last few years, many new approaches have been proposed which are elaborations ....

L.P. Kaelbling. An architecture for intelligent reactive systems. In Reasoning about actions and plans: Procedings of the

....component solutions into a smoothly functioning whole when they been developed independently. Hayes Roth states that agent architectures need to emphasize integration and interoperation. There is much work on agent architectures. Early architectures were created by Brooks [5] and Kaelbling [21]. More recent examples include LOOM [25] a language for modeling agents; Hayes Roth and Larsson s domain specific agent architecture [16] for medical monitoring tasks; Harandi and Rendon s intermediate level architecture [15] which provides rich, yet general purpose, support; and Sloman s global ....

# L.P. Kaelbling, An Architecture for Intelligent Reactive Systems, Morgan Kaufman, pp. 395-410, 1987.

....its expected position, a new object may block a pre computed motion, or other such problems may appear. In addition, an approximate model of the world can cause errors in the plan from which the robot will need to recover. The field of reactive planning [Schoppers, 1987; Agre and Chapman, 1987; Kaelbling, 1986] has concentrated on the problem of handling unanticipated events at execution time, but with only a limited ability to build plans. More recently, there has been an increasing effort to combine classical planning with reactive execution components [Gat, 1992; Lyons and Hendriks, 1992; Drummond ....

L. Kaelbling. An architecture for intelligent reactive systems. In Reasoning About Actions and Plans: Proceedings of the 1986 Workshop. AAAI, 1986.

.... by Pnueli and followers (see, e.g. Pnueli, 1986] and the description above) Second, researchers in AI planning take a reactive system to be one that is capable of responding rapidly to changes in its environment here the word reactive is taken to be synonymous with responsive (see, e.g. [Kaelbling, 1986]) More recently, the term has been used to denote systems which respond directly to the world, rather than reason explicitly about it (see, e.g. Connah and Wavish, 1990] In this paper the term is used in its Pnuelian sense, except where otherwise indicated. 2 In mainstream computer science, ....

....[Booch, 1994] gives a good discussion of software complexity and the role that object oriented development has to play in overcoming it. Russell and Norvig, 1995] introduced the ve point classi cation of environments that we reviewed here, and distinguished between the easy and hard cases. [Kaelbling, 1986] touches on many of the issues discussed here, and [Jennings, 1999] also discusses the issues associated with complexity and agents. 20 The formal model of agents and environments is similar to many that are now used in arti cial intelligence: Fagin et al. 1995] and [Genesereth and Nilsson, ....

Kaelbling, L. P. (1986). An architecture for intelligent reactive systems. In George, M. P. and Lansky, A. L., editors, Reasoning About Actions & Plans | Proceedings of the 1986 Workshop, pages 395-410. Morgan Kaufmann Publishers: San Mateo, CA.

.... control mechanisms are implemented, for instance, by classical planners [ Fikes and Nilsson, 1971; Wilkins, 1988 ] case based planners [ Hammond, 1990 ] conditional planners [ Schoppers, 1987 ] deferred planning techniques [ Durfee and Lesser, 1986 ] reactive systems [ Brooks, 1986; Kaelbling, 1987 ] Even though limited tuning operations are possible, most of the existing planning systems provide only one basic planning activity for a planning task and a fixed overall control mechanism. But real world and complex applications require that planning systems be customized flexibly according ....

L.P. Kaelbling. An architecture for intelligent reactive systems. In Reasoning about actions and plans: Procedings of the 1986 Workshop. Morgan-Kaufmann Publishers, 1987.

....to integrate the different level modules. Our hierarchical approach is different from a conventional hierarchical planner in the sense that each of the levels have access to input and output. Since it is possible to have conflicting output commands, we need some type of coordination or mediation [23] among them. Adopting the subsumption architecture s method of mediation, the outputs are suppressed by a higher level when the higher level makes an overriding decision. In the original version of the subsumption architecture, a time period is specified, during which the output will be ....

L. P. Kaelbling. An architecture for intelligent reactive systems. In Reasoning About Actions and Plans, pages 395 -- 410. Morgan Kaufmann, Los Altos, CA, 1987.

....of actions that leads from one specified configuration to another [FN71, AHT90] A fundamental question about planning is whether a substantial search is needed to find such a chain. Because of the computational costs of generic searching many approaches to avoiding it have been suggested (e.g. [HRHR79, Cha87, Kae87]) We suspect that the commonsense planning functions that biological systems perform can be realized with a KB in which statements specify the next step to be taken in most naturally arising situations, and therefore that extended searches can indeed be avoided. As an example, consider the case ....

L. Kaelbling. An architecture for intelligent reactive systems. In M. Georgeff and A. Lansky, editors, Reasoning about Actions and Plans, Los Altos, CA, 1987. Morgan Kaufmann.

....(Minton, Bresina, Drummond 1991) derives macro operators to reduce the depth of search. Each of these methods has the disadvantage of saving an indeterminate amount of plan generation time, and hierarchical planning can actually degrade the quality of the generated plan. Reactive planning (Kaelbling 1987; Brooks 1991) which limits the depth of the search space, can be significantly faster than nonreactive, or deliberative methods, but may produce suboptimal plans. Anytime planning (Boddy 1991; Zilberstein Russell 1996) allows a tradeoff between the solution quality of deliberative planning ....

Kaelbling, L. 1987. An architecture for intelligent reactive systems. San Mateo, CA: M. Kaufmann. Edited by Michael P. Georgeff and Amy L. Lansky.

....actions are imprecise and may fail. Classical planning approaches to the control of mobile robots have been criticized for not being able to adequately cope with uncertainty, imprecision and Currently at: Iridia, Universit e Libre de Bruxelles, Belgium, asaffio ulb.ac. be execution errors [4, 7, 5, 12]. To overcome these limitations, some authors have proposed architectures centered around a complex controller, or reactive planner (e.g. 4, 7] These reactive architectures provide immediate response to unpredicted environmental situations by giving up the idea of reasoning about future ....

....cope with uncertainty, imprecision and Currently at: Iridia, Universit e Libre de Bruxelles, Belgium, asaffio ulb.ac. be execution errors [4, 7, 5, 12] To overcome these limitations, some authors have proposed architectures centered around a complex controller, or reactive planner (e.g. [4, 7]) These reactive architectures provide immediate response to unpredicted environmental situations by giving up the idea of reasoning about future consequences of actions. Reasoning about future consequences, however, is still needed in order to intelligently solve complex tasks (e.g. by deciding ....

L. P. Kaelbling. An architecture for intelligent reactive systems. In M.P. Georgeff and A.L. Lansky, editors, Reasoning about Actions and Plans. Morgan Kaufmann, 1987.

....new information about goals and data during the plan execution. In order to avoid the computational cost of control reasoning and thereby be compatible with real time conditions, a number of researchers have turned their attention to the theory, design and implementation of reactive agents [9] [10], 11] Reactivity is an essential behavior, and a good architecture must support it. However, these systems do not seem to be appropriate for complex applications. They are difficult to design since knowledge is mainly expressed with perception action rules, and since all the possible situations ....

L. P. Kaelbling. An architecture for intelligent reactive systems. Reasoning about actions and plans. Morgan Kaufmann, 1987.

....PRS s ability to eliminate dead reckoning from lower abstraction levels was instructive, and on subsequent analysis pointed to the importance of having actions whose duration depends on the sensed environment, and to a roboticsoriented view of motion. The REX project (Rosenschein 1985; Kaelbling 1986) was equally influential: the behavior of a situated automaton is always contingent on the state of the environment. Indeed, the contingency assumes precisely the forms I have adopted for universal plans, with continuously evaluated predicates determining which numeric feedback function should be ....

L. Kaelbling. An architecture for intelligent reactive systems. Proc Workshop on Planning and Reasoning about Action, 1986, AAAI.

....issues. 1 Introduction A common assumption in early AI planning research was that in order to achieve a goal j, an agent should first compute an entire plan p for j, and then execute p [11] Within the AI planning community, this strategy has long been recognised to be a severe oversimplification [15]. In the real world, the assumptions that underpin plan execution are continually made false, either through the interference of other agents, or else by actions simply failing to have their intended effects. Recognition of this led to work on planning systems that could integrate planning, ....

L. P. Kaelbling. An architecture for intelligent reactive systems. In M. P. Georgeff and A. L. Lansky, editors, Reasoning About Actions & Plans --- Proceedings of the 1986 Workshop, pages 395--410. Morgan Kaufmann Publishers: San Mateo, CA, 1986.

....circuit, fast and without reason or explanation. Computational models of interaction like Brooks subsumption are fine for the tasks to which they are applied. However, they do not provide a uniform model of interaction at various cognitive levels. 2. 4 Reacting as an Automaton Kaelbling [ Kaelbling, 1986, Kaelbling, 1988 ] describes a reactive architecture for robot planning. This architecture defines a hierarchical organization of robot behaviors and a control scheme that determines appropriateness of available behaviors. Rosenschein s and Kaelbling s work [ Kaelbling, 1988, Kaelbling ....

Kaelbling, L. (1986). An architecture for intelligent reactive systems. reasoning about actions and plans. In M. G. nd A. Lansky (Ed.), Reasoning about Actions and Plans. Morgan Kauffman.

....The simple behaviors of a reactive control system acquire the information about the environment directly from the sensors, instead of getting this information indirectly through an intervening world model. These behaviors are closely tied to the effectors that carry out the behavior of the robot [12, 11]. These non representational systems avoid many of the pitfalls experienced by the traditional symbolic based and worldmodel driven systems, but they do have their own problems. The subsumption architecture does not allow for the explicit representation of high level goals, so it is difficult to ....

L. Kaelbling. An Architecture for Intelligent Reactive Systems. SRI Technical Note V 400, SRI International. Oct. 1986.

No context found.

L.P. Kaelbling. An architecture for intelligent reactive systems. In Workshop on Planning & Reasoning about Action, pages 235-250, Timberline OR, 1986.

No context found.

Kaelbling, L. P. (1990). An architecture for intelligent reactive systems. In James Allen, J. H. and Tate, A., editors, Readings in Planning, pages 713-728. Morgan Kaufmann.

No context found.

L. P. Kaelbling, An Architecture for Intelligent Reactive System, Reasoning about actions and plans: Proc. 1986 Workshop, pp. 395-410 Morgan Kauffman, (1987).

No context found.

L. P. Kaelbling, An Architecture for Intelligent Reactive System, Reasoning about actions and plans: Proc. 1986 Workshop, pp. 395-410 Morgan Kauffman, (1987).

No context found.

Kaelbling, L.P. (1990) An Architecture for Intelligent Reactive Systems. In: Readings in Plannings, pp. 713728.

No context found.

L. P. Kaelbling. An architecture for intelligent reactive systems. In J. Allen, J. Hendler, and A. Tate, editors, Readings in Planning, pages 713-728. Morgan Kaufmann publishers Inc.: San Mateo, CA, USA, 1990.

No context found.

L.P. Kaelbling. An architecture for intelligent reactive systems. In Reasoning about actions and plans: Procedings of the

No context found.

Kaelbling, L. P., "Architecture for Intelligent Reactive Systems", Proc. of the 1986 Workshop on Reasoning about Actions and Plans, Morgan Kaufmann, San Mateo, California, pp. 395-411, 1986.

No context found.

L. Kaelbling, "An architecture for intelligent reactive systems," in Proceedings of the Workshop on Planning & Reasoning about Action , AAAI, (1986).

No context found.

Kaelbling, L. An Architecture for Intelligent Reactive Systems. Technical report, SRI International, Menlo Park CA, 1987.

No context found.

Kaelbling, L. P., An Architecture for Intelligent Reactive Systems. In Reasoning about Actions and Plans, Lansky and Georgeff eds., pp. 395-410, Morgan Kaufman, 1987.

First 50 documents  Next 50

Online articles have much greater impact   More about CiteSeer.IST   Add search form to your site   Submit documents   Feedback  

CiteSeer.IST - Copyright Penn State and NEC