Suppose one is given a description of a system, together with an observation of the system's behaviour which conflicts with the way the system is meant to behave. The diagnostic problem is to determine those components of the system which, when assumed to be functioning abnormally, will explain the discrepancy between the observed and correct system behaviour. We propose a general theory for this problem. The theory requires only that the system be described in a suitable logic. Moreover, there are many such suitable logics, e.g. first-order, temporal, dynamic, etc. As a result, the theory accommodates diagnostic reasoning in a wide variety of practical settings, including digital and analogue circuits, medicine, and database updates. The theory leads to an algorithm for computing all diagnoses, and to various results concerning principles of measurement for discriminating among competing diagnoses. Finally, the theory reveals close connections between diagnostic reasoning and nonmonotonic reasoning.

Qualitative simulation predicts the set of possible behaviors...

Renewed motives for space exploration have inspired NASA to work toward the goal of establishing a virtual presence in space, through heterogeneous effets of robotic explorers. Information technology, and Artificial Intelligence in particular, will play a central role in this endeavor by endowing these explorers with a form of computational intelligence that we call remote agents. In this paper we describe the Remote Agent, a specific autonomous agent architecture based on the principles of model-based programming, on-board deduction and search, and goal-directed closed-loop commanding, that takes a significant step toward enabling this future. This architecture addresses the unique characteristics of the spacecraft domain that require highly reliable autonomous operations over long periods of time with tight deadlines, resource constraints, and concurrent activity among tightly coupled subsystems. The Remote Agent integrates constraint-based temporal planning and scheduling, robust multi-threaded execution, and model-based mode identification and reconfiguration. The demonstration of the integrated system as an on-board controller for Deep Space One, NASA's rst New Millennium mission, is scheduled for a period of a week in late 1998. The development of the Remote Agent also provided the opportunity to reassess some of AI's conventional wisdom about the challenges of implementing embedded systems, tractable reasoning, and knowledge representation. We discuss these issues, and our often contrary experiences, throughout the paper.

Industrial process plants such as chemical refineries and electric power generation are examples of continuous-variable dynamic systems (CVDS) whose operation is continuously monitored for abnormal behavior. CVDSs pose a challenging diagnostic problem in which values are continuous (not discrete), relatively few parameters are observable, parameter values keep changing, and diagnosis must be performed while the system operates. We present a novel method for monitoring CVDSs which exploits the system's dynamic behavior for diagnostic clues. The key techniques are: modeling the physical system with dynamic qualitative /quantitative models, inducing diagnostic knowledge from qualitative simulations, continuously comparing observations against fault-model predictions, and incrementally creating and testing multiple-fault hypotheses. The important result is that the diagnosis is refined as the physical system's dynamic behavior is revealed over time. Introduction Process monitoring is a c...

Incomplete knowledge of the structure of mechanisms is an important fact of life in reasoning, commonsense or expert, about the physical world. Qualitative simulation captures an important kind of incomplete, ordinal, knowledge, and predicts the set of qualitatively possible behaviors of a mechanism, given a qualitative description of its structure and initial state. However, one frequently has quantitative knowledge as well as qualitative, though seldom enough to specify a numerical simulation.

This paper investigates two different activities that involve making assumptions: predicting what one expects to be true and explaining observations. In a companion paper, an architecture for both prediction and explanation is proposed and an implementation is outlined. In this paper, we show how such a hypothetical reasoning system can be used to solve recognition, diagnostic and prediction problems. As part of this is the assumption that the default reasoner must be "programmed" to get the right answer and it is not just a matter of "stating what is true" and hoping the system will magically find the right answer. A number of distinctions have been found in practice to be important: between predicting whether something is expected to be true versus explaining why it is true; and between conventional defaults (assumptions as a communication convention), normality defaults (assumed for expediency) and conjectures (assumed only if there is evidence). The effects of these distinctions on...

Efficient implementations of DPLL with the addition of clause learning are the fastest complete Boolean satisfiability solvers and can handle many significant real-world problems, such as verification, planning and design. Despite its importance, little is known of the ultimate strengths and limitations of the technique. This paper presents the first precise characterization of clause learning as a proof system (CL), and begins the task of understanding its power by relating it to the well-studied resolution proof system. In particular, we show that with a new learning scheme, CL can provide exponentially shorter proofs than many proper refinements of general resolution (RES) satisfying a natural property. These include regular and Davis-Putnam resolution, which are already known to be much stronger than ordinary DPLL. We also show that a slight variant of CL with unlimited restarts is as powerful as RES itself. Translating these analytical results to practice, however, presents a challenge because of the nondeterministic nature of clause learning algorithms. We propose a novel way of exploiting the underlying problem structure, in the form of a high level problem description such as a graph or PDDL specification, to guide clause learning algorithms toward faster solutions. We show that this leads to exponential speed-ups on grid and randomized pebbling problems, as well as substantial improvements on certain ordering formulas. 1.

This paper develops the Reactive Model-Based Programming Language (RMPL) and its executive, called Titan. RMPL provides the features of synchronous, reactive languages, with the added ability of reading and writing to state variables that are hidden within the physical plant being controlled. Titan executes an RMPL program using extensive component-based declarative models of the plant to track states, analyze anomalous situations, and generate novel control sequences. Within its reactive control loop, Titan employs propositional inference to deduce the system's current and desired states, and it employs model-based reactive planning to move the plant from the current to the desired state

If one wants to use logic to build a diagnostic system, then it is not a matter of "just axiomatising" the domain; we have to understand how to use logic for diagnosis. We need some models of what diagnosis is, in order to be able to implement diagnostic systems. This paper considers 3 different "logical " definitions of diagnosis. Each of these are presented in a uniform framework of hypothetical reasoning where the user provides the possible hypotheses. These are compared as to the sort of knowledge that we need to provide them, and in their expressibilty. It seems as though there is no one framework which can claim to be the logical definition of diagnosis. Each of these approaches has been implemented in the Theorist system, and used on a number of domains. This paper concentrates on the case where we have fault models. 1 Introduction Diagnosis is a problem of trying to find what is wrong with some system based on knowledge about the design /structure of the system, possible malf...

Artificial intelligence has traditionally used constraint satisfaction and logic to frame a wide range of problems, including planning, diagnosis, cognitive robotics and embedded systems control. However, many decision making problems are now being re-framed as optimization problems, involving a search over a discrete space for the best solution that satisfies a set of constraints. The best methods for finding optimal solutions, such as A*, explore the space of solutions one state at a time. This paper introduces conflict-directed A*, a method for solving optimal constraint satisfaction problems. Conflict-directed A* searches the state space in best first order, but accelerates the search process by eliminating subspaces around each state that are inconsistent. This elimination process builds upon the concepts of conflict and kernel diagnosis used in model-based diagnosis[1,2] and in dependency-directed search[3--6]. Conflict-directed A* is a fundamental tool for building model-based embedded systems, and has been used to solve a range of problems, including fault isolation[1], diagnosis[7], mode estimation and repair[8], model-compilation[9] and model-based programming[10].