Erik Sandewall, Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994  Home/Search   Document Not in Database   Summary   Related Articles   Check

....problem. The underlying principles of our theory, however, are su# ciently fundamental and general to not depend on this specific language. Thus these principles could equally well be employed in other, more elaborated formal theories of actions like, e.g. Gelfond and Lifschitz, 1993; Sandewall, 1994; Thielscher, 1995 ] in view of the qualification problem. Likewise, existing action calculi may be enhanced on this basis in order that they become capable of dealing with abnormal action disqualifications. As an example, we have sketched a way to embed the fluent calculus in an appropriate ....

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994.

....may serve as a high level surface language for specifying action domains in FLUX. 1 Introduction An unsatisfactory aspect of research into reasoning about actions is the coexistence of a variety of di#erent approaches, which are di#cult to assess and compare. Systematic approaches such as [11] or the Action Description Language [2] have been developed to help eliminate this deficiency. Providing a high level but formal semantics, these approaches are intended to be used to prove correctness of di#erent action calculi for well defined problem classes. A formal evaluation is ....

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994.

....or require invocation of other agents actions. The distinction between structural ramifications when the action can affect features of other objects than those which occur as arguments of the action and local ramifications involving only features of the argument objects was identified in [Sandewall, 1994]. For example, the following domain constraint H(t, near(x) y) H(t, near(y) x) 3 Tactical Agent Architecture 3.1 Task Oriented Agents The behaviour functions of the situated agents, described in the previous section, are uniformly defined across their respective domains and ranges. This ....

....channel C Figure 1. Tactical Agent Architecture Tactical Coordination A dynamical system can be characterised as a system whose state changes over time, and where effects flow forward in time so that the non input part of the state at one time can only depend on its earlier states [Sandewall, 1994]. Agents of the system perform actions influencing state variables and changing the system state. We define a dynamic multi agent system by a set of architecture types A c As, At, An, AH, Aro, Aro, Avo, Avo , and a particular value of a time parameter t. Given a finite set of agents gk (1 ....

Sandewall, E. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press 1994.

.... for reasoning about actions, notably the situation [19, 16] the uent [11, 31] and the event calculus [13, 27] approaches based on the linear connection method [1, 2] linear logic [8, 17] transaction logic [3] temporal action logics [4] action languages [7] the features and uent approach [26], etc. We opt for the uent calculus because it is a logic with standard semantics and a sound and complete calculus, and many of the problems in reasoning about actions like the frame and the rami cation problem can be dealt with in a representationally as well as computationally adequate way. ....

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994.

.... of proposals available this days, notably the situation [19, 15] uent [11, 24] and event calculus [14, 22] approaches based on the linear connection method [1, 2] linear logic [7, 17] transaction logic [3] temporal action logics [5] action languages [6] the features and uent approach [21], etc. The most prominent problem addressed in these approaches is the problem of whether a certain goal can be reached by executing a certain sequence of actions in a certain situation. This is an instance of the well known entailment problem. This paper deals with the entailment problem in two ....

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994.

....state descriptions have been restricted, for the sake of simplicity, to propositional constants throughout the first part of the paper. For our calculus, we introduce a richer notion of fluents. A fluent is now an n place predicate with arguments chosen from a given set of objects (or entities) Sandewall, 1994; Kartha and Lifschitz, 1994 ] This involves both a generalized concept of action laws and fluent formulas including quantifications. Definition 15 Let E be a finite set of symbols called entities . Let F denote a set of fluent names, each of which is associated with a natural number called ....

....problem. The underlying principles of our theory, however, are sufficiently fundamental and general to not depend on this specific language. Thus these principles could equally well be employed in other, more elaborated formal theories of actions like, e.g. Gelfond and Lifschitz, 1993; 34 Sandewall, 1994; Thielscher, 1995 ] to tackle the qualification problem. Likewise, existing action calculi may be enhanced on this basis in order that they become capable of dealing with abnormal action disqualifications. As an example formalism, in the second part of the paper we have embedded the fluent ....

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994.

....this solution to the frame problem is thus the splitting of the theory into different parts, which are circumscribed separately. This technique, sometimes referred to as forced circumscription, is also employed in [CrEt 92] Dohe 94] and [KaLi 95] and is akin to what Sandewall calls filtering [Sand 94] 2.3 Narrative Information and Planning In some circumstances it is convenient to define Happens in terms of other predicates representing different categories of action occurrence. For 6 example, in the context of planning we may wish to distinguish between actions that have (definitely) ....

....this section we illustrate how this facility for dynamic management of the frame can be incorporated into the Event Calculus by use of a new predicate Releases A Theta F Theta T . A form of this predicate was first introduced in [KaLi 95] and it is related to Sandewall s idea of occlusion [Sand 94] Releases(A; F; T ) expresses that if A occurs at T it will disable the fluent F s innate persistence. The truth value of F will then be free to fluctuate until the next action occurrence which initiates or terminates it. Releases is defined in the domain dependent part of the theory and ....

E. Sandewall, The Representation of Knowledge about Dynamical Systems, Volume 1, Oxford University Press, 1994.

....states can be represented as homogeneous and persistent concepts. As a motivation for introducing the possibilityofrepresenting persistent properties in the language, this Section considers how to solve the frame problem, and in particular the famous example of the Yale Turkey Shooting Scenario #Sandewall, 1994; Allen Ferguson, 1994#, formerly known as the Yale Shooting Problem. An inertia operator #= is introduced here. Intuitively, C is currently true if it was true at a preceding interval # say i # and there is no evidence of the falsity of C at any interval between the current one and i. Thus, ....

Sandewall, E. #1994#. Features and Fluents. The Representation of Knowledge about Dynamical Systems,Vol. I. Oxford University Press.

....decisions for making precise the various planning tasks and their solutions, and heuristic search algorithms for obtaining those solutions. The use of high level representation languages is common in AI Planning [25, 50] and our approach draws insight from recent work on theories of action [29, 54, 28]. Similarly, the use of mathematical models of sequential decisions such as Markov Decision Processes (MDPs) and Partially Observable MDPs (POMDPs) borrows from the work in Dynamic Programming [51, 6] Finally, the use heuristic search algorithms has long tradition in AI (e.g. 46] even though ....

....that illustrate the scope of the approach and the capabilities of the tool. The planning approach described is a natural integration of a number of ideas from AI and Dynamic Programming: State models, mdps, and pomdps, 4, 57, 17] rtdp algorithms [39, 3] and action representation languages [29, 52, 54]. At the same time, it s related to a number of decision theoretic approaches to planning such as [41, 24, 14] while being distinct in the use of a more expressive logical language for describing actions and sensors, and the use of the rtdp algorithm for solving a variety of decision models. The ....

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford Univ. Press, 1994.

....system 3 2 Components of the system We propose then the architecture of the Decision Support System which will help deciders to achieve those goals in figure 2. In general we can define a system as a collection of objects which have individual properties as well as relationships between each other [22]. We want to build the system of figure 2. We will then in the following lines describe the components of the system. 2.1 Knowledge Based System: Every action of users and behaviour of nature has consequences. The system has to generate from a situation the corresponding consequences. We have then ....

....not only the direct consequences of the usage of the resources but also consequences of those consequences, and the consequences due to the behaviour of nature. 2. 3 Physical dynamics of the system The real world is a dynamic system, because properties of its objects change according to time [22] but space has also to be take into account. The goal is to build a system which will simulate the behaviour of resources and users depending on space and time. 2.3.1 Space We consider the problem of using a resource, as the one of accessing the space in which this resource belongs. Then the ....

E. Sandewall. Features and Fluents, The Representation of Knowledge about Dynamical Systems, volume 1. Oxford Science Publications, 1994.

....10 percent of the full algebra. None of these algebras, however, are capable of expressing the notion of sequentiality, which is that of specifying that some intervals have to occur in sequence in time, without any overlap. This is required e.g. in some cases of reasoning about action [8]. The maximality result of the ORD Horn algebra then implies that the requirement that the algebra contain all the basic relations has to be sacrificed. Golumbic and Shamir [5] come close to expressing sequentiality, but require that any two intervals are related. In this paper, we exploit a ....

.... the algebras using the utility aclose [6] 2 We have four algebras A(r; OE) all containing the relations (j) OE) OE j) j) and (OE ) expressing the notion of sequentiality, which is useful for solving reasoning problems under the assumption that actions always occur in sequence [8]. Note that the ORD Horn algebra does not contain the relation (OE ) and thus cannot express sequentiality. 10 Proposition 27 The eight algebras A(r; b) which have b 2 ff; sg are maximal tractable algebras. Proof: By running the utility atry [6] which generates minimal extensions of ....

[Article contains additional citation context not shown here]

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems, volume I. Oxford University Press, 1994. ISBN 0-19-853845-6.

....invocation of other agents actions. The distinction be tween structural ramifications when the action can affect features of other objects than those which occur as arguments of the action and local ramifications involving only features of the argument objects was identified by Sandewall [12]. For example, the following domain constraint H(t, near(x) y) H(t, near(y) x) demands from a model to include the atomic formula near(B) A, whenever it contains the atomic formula near(A) B. Therefore, at the moment when agent A evaluates near(A) B as true (either by sensing a new ....

....to simple reactions. 7 Figure 1. Tacitcat Agent Arctctre 4 Tactical Coordination A dynamical system can be characterised as a system whose state changes over time, and where effects flow forward in time so that the non input part of the state at one time can only depend on its earlier states [12]. Agents of the system perform actions influencing state variables and changing the system state. We define a dynamic multi agent system by a set of architecture types A C As, AT, AH, AEH, ATe, AETe, Ape, AEpe , and a particular value of a time parameter t. Given a finite set of agents gk (1 ....

Erik Sandewall. Features and Fluents. The Representation of Knowl- edge about Dynamical Systems. Volume I. Oxford University Press 1994.

....Abstract. Explanation closure (EC) axioms were previously introduced as a means of solving the frame problem. This paper provides a thorough demonstration of the power of EC combined with action closure (AC) for reasoning about dynamic worlds, by way of Sandewall s test suite of 12 or so problems [29 31]. Sandewall s problems range from the Yale turkey shoot (and variants) to the stu y room problem, and were intended as a test and challenge for nonmonotonic logics of action. The EC AC based solutions for the most part do not resort to nonmonotonic reasoning at all, yet yield the intuitively ....

....no inferences based on world knowledge are deductive, whether directed forward or backward in time. 4 The usefulness of EC axioms in planning has also become apparent in more recent work on SAT planning (e.g. 14] 5 These publications were precursors of the monograph Features and Fluents[31]. Lenhart Schubert The test suite provides an unprecedented opportunity to examine the strengths and shortcomings of various methods for reasoning about change in a systematic way. I will show that the approach based on EC reasoning fares very well indeed. Moreover, the pro ered solutions are ....

[Article contains additional citation context not shown here]

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

....with strict inertia (no qualification, ramification, concurrency, or surprises) but allowing nondeterministic actions and actions with extended duration in time. They have been reported in conference, journal, and handbook articles [7, 9, 10] and in a monograph that has appeared very recently [8]. The development of the coherent framework for the assessment oriented approach has been a major part of that work. It included: ffl The definition of the underlying semantics. Our choice of underlying semantics is based on the use of an ego world game, that is, a formalized version of the ....

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

....authors, and which differ in their details. So far, these methods have only been motivated on the basis of (1) intuitive and informal reasons why a particular method is the natural one, and (2) selected test examples where competing methods do not give intended results. In earlier articles [17, 18] we have argued that this kind of evidence is not sufficient, and that proposed methods for nonmonotonic common sense reasoning should be assessed with respect to an underlying semantics which formally defines the intended conclusions for a class of scenario descriptions. We have also performed ....

....functions and the corresponding action laws, the problem of non initial observations which lead to postdiction problems, etc. Fortunately, however, all of these problems have been adequately handled already in the context of strict inertia, and can be considered as routine matters. We refer to [18] for all the necessary details. Therefore, we can proceed entirely on the semantic level here. 5 MEASURE SYSTEMS We proceed now to the analysis of cases where straight ramification is not sufficient. For our analysis of the commonly used, minimization based approaches, we introduce the following ....

[Article contains additional citation context not shown here]

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

....authors, and which differ in their details. So far, these methods have only been motivated on the basis of (1) intuitive and informal reasons why a particular method is the natural one, and (2) selected test examples where competing methods do not give intended results. In earlier articles [San93, San94] we have argued that this kind of support is not sufficient, and that proposed methods for nonmonotonic commonsense reasoning should be assessed with respect to an underlying semantics which formally defines the intended conclusions for a class of scenario descriptions. We have also performed such ....

....using logic formulae. In the present work, we obtain a much more concise analysis of the problem by expressing it in terms of next state functions instead. The systematic translation from such next state functions to corresponding action laws is a routine matter, which has been shown and used in [San94]. The idea is simple. Let R be the set of possible states of the world, for example, the Cartesian product of the range sets of the respective state variables, and let R c be a subset of R consisting of those states that satisfy the static domain constraints. Also, let E be the set of possible ....

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

No context found.

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems, volume 1 of Oxford Logic Guides. Oxford University Press, 1994.

No context found.

Erik Sandewall, Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994

No context found.

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Oxford University Press, 1994.

No context found.

E. Sandewall. Features and Fluents. The representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

No context found.

E. Sandewall, Features and Fluents, The Representation of Knowledge about Dynamical Systems, Vol. I, Oxford University Press, 1994.

No context found.

E. Sandewall. Features and Fluents. The representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

No context found.

Erik Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Volume I. Oxford University Press, 1994.

No context found.

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Vol. 1. Oxford University Press, 1994.

No context found.

E. Sandewall. Features and Fluents. The Representation of Knowledge about Dynamical Systems. Vol. 1. Oxford University Press, 1994.

First 50 documents

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