By Topic

AI, Simulation and Planning in High Autonomy Systems, 1991. Integrating Qualitative and Quantitative System Knowledge, Proceedings of the Second Annual Conference on

Date 1-2 April 1991

Filter Results

Displaying Results 1 - 25 of 38
  • Proceedings of the Second Annual Conference on AI, Simulation and Planning in High Autonomy Systems. Theme: Integrating Qualitative and Quantitative System Knowledge

    Save to Project icon | Request Permissions | PDF file iconPDF (1391 KB)  
    Freely Available from IEEE
  • Abstraction morphisms for task planning and execution

    Page(s): 50 - 59
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (752 KB)  

    Deals with further efforts to develop abstraction mechanisms for systematic derivation of related models through the use of system morphisms. The authors describe an abstraction mechanism for mapping a task plan hierarchy into an isomorphic tree of model abstractions that supports hierarchical task execution. The task plan hierarchy is formulated in a model-based planning approach. They also show how the hierarchical execution structure can be constructed. Then the endomorphism concept employed in the modelling of autonomous systems is illustrated. The DEVS-Scheme knowledge-based, discrete event simulation environment is used to test the models and tools in an autonomous laboratory application View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Model-based troubleshooting of complex technical systems using integrated qualitative techniques

    Page(s): 122 - 129
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (572 KB)  

    Monitoring and diagnosis of safety-critical dynamic systems are challenging tasks for knowledge-based programming. While current control systems usually rely on the total availability and computability of numerical information, they necessarily fail if these assumptions do not hold. As a consequence. Intelligent control systems are needed which have the ability to perform monitoring and diagnosis even if only incomplete or abstract pieces of information are available. Additionally, monitoring and diagnosis have to be integrated into an overall online control cycle in order to apply the concepts to dynamic systems which are characterized by time-varying parameters. An implemented expert system which fulfils these requirements is presented. Research work done in the field of model-based diagnosis of nuclear power plants using qualitative reasoning techniques is given View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Symbolic discrete event system specification

    Page(s): 130 - 141
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (804 KB)  

    Extending discrete event modelling formalisms to facilitate greater symbol manipulation capabilities is important to further their use in intelligent control and design of high autonomy systems. This paper defines an extension to the discrete event system specification (DEVS) formalism that facilitates symbolic expression of event times by extending the time base from the real numbers to the field of linear polynomials over the reals. A simulation algorithm is developed to generate the branching trajectories resulting from the underlying nondeterminism. The extended formalism offers a convenient means to conduct multiple, simultaneous explorations of model behaviors. Examples of application are given with concentration on fault model analysis View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Dynamic selection of models under time constraints

    Page(s): 60 - 67
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (640 KB)  

    The creation of an appropriate model to use for a specific task involves determining the appropriate assumptions to simplify a more complex base model. The author presents a conceptual framework for organizing a series of models that are derived with simplifying assumptions, and a decision-theoretic model-based method to determine the optimal model to select under a time constraint. He then discusses a heuristic approach to the dynamic selection of models under time constraints. He illustrates this approach by describing how it can be applied to an application to reason about physiologic abnormalities of patients in the intensive care unit (ICU) who are being treated with a mechanical breathing device (ventilator) View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Planning with events and states

    Page(s): 181 - 186
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (432 KB)  

    Presents an overall planning system in which specifications can be described in terms of events and states. The underlying feature of this system is temporal logic, and its expressive power alloys one to deal with simultaneous actions and interacting actions. Moreover, one can represent both goal-oriented positive constraints and prevention-oriented negative constraints. The planning system can generate hierarchical plans and the overall model is capable of handling interacting agents View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Temporal information in qualitative simulation

    Page(s): 298 - 305
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (564 KB)  

    Qualitative simulation is a valuable method for predicting the behavior of partially known dynamical systems. This paper presents an order of magnitude reasoning method which combines purely qualitative simulation technique with some partial numerical information in order to capture temporal information. Duration evaluation methods used so far are based on the mean value theorem. However, for critical points, this method is no longer suitable and a method based on the second order Taylor formula is proposed and presented with a validating example View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Planning by autonomous agents with many concurrent goals in an elaborate simulated world

    Page(s): 15 - 19
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (424 KB)  

    Complex simulated worlds populated by simulated agents having many concurrent goals present unique problems to planning systems. Neither a purely reactive model nor a traditional AI planning model is entirely adequate. The first because the agents need to know what they will be doing in the future to make good decisions about what to do in the present. The second because of the size and scope of the problems presented. The author describes a multi-level hybrid planning system consisting of a pre-formed plan language, a conflict resolution at the action level mechanism, and a coarse-grained, plan activities for the future facility that enables the Main Street simulated world to react to highly diverse sets of circumstances as well as to engage in long-range planning for the future View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Representing temporal spatial, and causal knowledge for monitoring and control in an Intelligent Simulation Training System

    Page(s): 263 - 269
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (452 KB)  

    The objective of the Intelligent Simulation Training System (ISTS) is to train students in monitoring and controlling physical objects in time and space. The ISTS consists of a graphic computer simulation, an expert system, and a user interface. The expert system must possess a high-level model for reasoning about simulation events. Quantitative simulation data and elicited domain expertise are used to generate an abstract, qualitative model for reasoning purposes. The qualitative model is generated by a subsystem known as ELICIT (Expertise Learner and Intelligent Causal Inference Tool). This paper describes the representations used for generating such a model. ELICIT is implemented in Common LISP and Joshua (a knowledge representation language developed by Symbolics Inc.) on a Symbolics 3630 LISP machine View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Diagnosability and sensor reduction

    Page(s): 142 - 146
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (384 KB)  

    Device-centered model-based reasoning (MBR) models are used to determine the diagnosability of individual components and of the system as a whole. Given a satisfactory device-centered model that can be used dependably for diagnosis, the author first considers the specification of diagnosability for individual components. This is expressed through uniqueness requirements for the isolation of faults in those components. The diagnosability of the system as a whole is a composite of the diagnosability of its components. This leads one to a criterion for determining the diagnostic value to the system of individual sensors. This criterion permits the maximization of system diagnosability with a fixed number of sensors or, conversely, minimizing the number of sensors required to achieve a given standard of system diagnosability View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • DEBORA: a decision engine based on rational aggregation

    Page(s): 68 - 77
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (796 KB)  

    Constraint propagation is a matter of logical deduction, but this is not always sufficient to reach a problem solution. Heuristic knowledge is usually needed to go on with the solution search when the first stage stops. The way this second type of knowledge is handled has more to do with decision rather than deduction process. In this paper the authors suggest a method to handle heuristic knowledge based on social choice theory. An analogy is proposed between the cooperation problem among heuristics expressed as decision rules and the voting problem. This analogy allows one to define and justify aggregation modes for results provided by each rule, with a view to providing a global decision ranking. An application to job-shop scheduling has been carried out View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Integrating multiple representations for incremental, causal simulation

    Page(s): 88 - 96
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (644 KB)  

    Many engineering problems require the ability to answer `what if' questions about the effects of complex physical events. To address such problems, the author has implemented MIDAS, a system that can incrementally simulate events and produce causal explanations describing their effects. MIDAS incorporates an explicit causal model of time, change, and persistence. It integrates multiple specialized representations using a truth maintenance system that records belief justifications and enables incremental assertion and retraction of beliefs. MIDAS has been tested in several complex domains, including geology and semiconductor fabrication View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Simulation-based planning of robot tasks in flexible manufacturing

    Page(s): 166 - 173
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (568 KB)  

    A framework is proposed for support of design, task planning, and simulation of automated manufacturing systems. The framework establishes a hierarchy of method banks essential for improving the efficiency and cost effectiveness of manufacturing processes. The methods should support automatic generation of sequencing rules, design and configuration of the manufacturing facility and equipment, synthesis of task oriented robot programs, and generation and execution of simulation models of a manufacturing system. In this paper, each layer is addressed and preliminary results that apply simulation to interpret and test task oriented robot programs are discussed View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Creating qualitative and combined models with discrete events

    Page(s): 306 - 315
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (660 KB)  

    Combined models with discrete event and continuous components can represent a wide variety of complex systems that must be defined using many different models. In artificial intelligence, similar efforts are underway in the study of qualitative models for physics and reasoning about systems. Consequently, there is a need to bridge theory and technology in order to have a uniform language when either analyzing or reasoning about physical systems. The authors present how combined modeling and discrete event modeling within the simulation literature can be used to help formulate such a bridge. They present formalisms from systems theory as well as the DEVS formalism to demonstrate the underlying mathematical foundation for talking about complex systems. They also present a precise methodology for partitioning continuous systems into discrete event systems for the purpose of defining qualitative models from quantitative models View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Actions as evidence: multiple epistemic agents acting under uncertainty

    Page(s): 21 - 30
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (632 KB)  

    Extends Ruspini's epistemic logic-based model for evidential reasoning by formalizing the actions of the epistemic agents and interpreting them as further sources of evidence. The actions of any intelligent agent, whether friend, foe, or neutral, provide evidence about the state of the world which is quite different from the kinds of evidence provided by a deterministic or stochastic natural process. The authors propose a new procedure for reasoning under uncertainty which combines the concepts of the theory of evidence, epistemic logic, and utility theory. The new procedure is applied to an example of a two intelligent agent problem where the second agent interprets the action of the first as evidence about probabilities of the possible worlds View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A qualitative circuit simulator

    Page(s): 248 - 252
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (408 KB)  

    The work described here is part of a larger project which is investigating various issues related to diagnosis and failure prediction in the context of automotive manufacturing. In particular, the goal is to produce prototype intelligent tools which will assist in the analysis of potential faults and hazard situations in evolving electrical circuit designs. The paper reports on the design and implementation of a qualitative electrical circuit simulator that can effectively model the structure and behaviour of systems under analysis. The simulator assigns labels to open and short circuit branches, finds current paths and determines directions of flow View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • ALCMEN: a language for qualitative/quantitative knowledge representation in expert supervisory process control

    Page(s): 80 - 87
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (380 KB)  

    ALCMEN (automaticians language for causal modelisation for expert knowledge) is a language capable to handle simultaneously imprecision and uncertainty, and precise equations, in order to solve the communication problem between process experts and control engineers. ALCMEN appears as a network of interconnected blocks and a list of structured variables. It captures the causal influence of a cause on an effect and provides the possibility of parametrizing that relation by one or many conditions. Each causal relation, is described according to the most explicit available knowledge, either using equations or natural language. Specification methodology is also provided using real time expert system development language G2 View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Extracting qualitative descriptions from quantitative models

    Page(s): 108 - 119
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (1032 KB)  

    Presents a general methodology for generating qualitative descriptions of specific physical systems from their quantitative model. Given a physical system, the methodology provides guidelines for formulating the system's quantitative model, and systematically exploring its parameter space to extract a qualitative description in the form of a graph. The qualitative features of the graph can increase an engineer's understanding of the physical system, and help with the design of an agent to control it. With suitable characterizations of its regions, the graph can also be used by the controlling agent itself. The methodology is applicable to an interesting domain of physical systems, and it is applied to an example problem from this domain-the two-dimensional baseball problem View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • A model-based approach for organizing quantitative computations

    Page(s): 210 - 218
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (652 KB)  

    Model based reasoning (MBR) is currently receiving wide spread attention because it offers a way to circumvent the brittleness of reasoning systems built solely on associational knowledge. To date, most MBR approaches have focused on the use and manipulation of qualitative models. The authors report their experience in applying techniques of functional reasoning to the general problem of organizing quantitative calculations. As a testbed, they have solved a problem initially posed at the Model-Based Diagnosis workshop (Paris, July, 1989): representing an automotive cruise control system. The results show that the principles of the functional reasoning approach can provide leverage in device domains characterized by quantitative data. A discussion of the current state of research in model based reasoning is included View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Reasoning about global behavior of ordinary differential equations by combining qualitative and quantitative analysis

    Page(s): 98 - 107
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (596 KB)  

    The authors attempt to integrate numerical methods and knowledge-based methods with qualitative reasoning as a kernel. The essence of the approach is threefold: (a) representing geometric and topological aspects of solution curves relevant to qualitative analysis as mappings between hyperplanes in the phase space; (b) computing mappings that characterize the behavior by local analysis of solution curves; and (c) deriving global behaviors by analyzing structural information of the composite mappings representing solution curve. Preliminary results obtained from this approach are demonstrated for two-dimensional ordinary differential equations View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Using qualitative knowledge for quantitative simulation of the human spatial orientation system

    Page(s): 279 - 288
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (728 KB)  

    Theory formation is an important scientific goal. This process often involves the development of complex mathematical models which are first qualitatively tested before being fine tuned quantitatively. The authors explore the quantitative fine tuning aspect of theory formation. The scientific domain chosen for this work is the human orientation system. The particular experiment for the modeling is space-borne. The scarceness of data leads to fine tune the model parameters for each astronaut. The authors are currently developing a computer tool for automatic, reliable and rapid convergence toward the best parameter fit for each individual subject using qualitative knowledge of the general model formulated View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • The use of multiple models in evaluating complex engineering designs

    Page(s): 225 - 232
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (552 KB)  

    The paper establishes a conceptual framework for viewing engineering design as a decomposition of the design problem into multiple models. Also described is a computational framework for iteratively refining designs through analysis and interpretation of candidate solutions. In doing so specific evaluation objectives and design descriptions are mapped to specific evaluation, analysis and interpretation models. Finally, the paper describes an implemented tool which uses the framework for evaluating a class of discrete event systems View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Sensor selection techniques in device monitoring

    Page(s): 154 - 163
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (720 KB)  

    For complex systems with large sensor complements, the authors develop a selective monitoring strategy to avoid information overload on system operators. They describe an approach to determining from moment to moment which subset of the available sensor data for a system is most informative about the state of the system and about interactions occurring within the system. They term this process sensor selection. The approach draws on concepts from causal reasoning and information theory. After describing the NASA test domain, they conclude with a report on the status of the implementation of the SELMON system View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Qualitative state spaces: a formalization of the naive physics approach to knowledge-based reasoning

    Page(s): 40 - 49
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (648 KB)  

    The authors attempt to formalize the naive physics approach to knowledge-based qualitative reasoning in such a way that it can be compared with the quantitative analysis techniques used in system theory. It is shown that naive physics models resemble classical quantitative models in more respects than is commonly assumed. This resemblance opens up an entire catalog of currently unanswered questions relating to naive physics models. It helps prove some theorems about such qualitative models, but it also unveils some of the shortcomings of these models View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.
  • Integration of qualitative and quantitative methods in visual reasoning

    Page(s): 272 - 278
    Save to Project icon | Request Permissions | Click to expandQuick Abstract | PDF file iconPDF (472 KB)  

    Describes how commonsense spatial reasoning is accomplished in a computational model that is primarily qualitative in nature, but which allows the smooth integration of quantitative methods as and when necessary. This model, termed visual reasoning, is characterized by representations that have symbolic and imaginal parts, visual operations that access spatial information contained in the imaginal parts, and visual cases which encode chunks of inferential knowledge. The authors have identified three conditions under which the invocation of quantitative or numerical methods become necessary in order for visual reasoning to proceed. These are described and illustrated by three examples of problem solving that show how quantitative methods get integrated into the framework of visual reasoning View full abstract»

    Full text access may be available. Click article title to sign in or learn about subscription options.