Mark S. Boddy

Deliberation scheduling for problem solving in time-constrained environments

Abstract We are interested in the problem faced by an agent with limited computational capabilities, embedded in a complex environment with other agents and processes not under its control. Careful management of computational resources is important for complex problem-solving tasks in which the time spent in decision making affects the quality of the responses generated by a system. This paper describes an approach to designing systems that are capable of taking their own computational resources into consideration during planning and problem solving. In particular, we address the design of systems that manage their computational resources by using expectations about the performance of decision-making procedures and preferences over the outcomes resulting from applying those procedures. Our approach is called deliberation scheduling. Deliberation scheduling involves the explicit allocation of computational resources to decision-making procedures based on the expected effect of those allocations on the systems performance.

Reasoning about partially ordered events

Abstract This paper describes a class of temporal reasoning problems involving events whose order is not completely known. We examine the complexity of such problems and show that for all but trivial cases these problems are likely to be intractable. As an alternative to a complete, but potentially exponential-time decision procedure, we provide a partial decision procedure that reports useful results and runs in polynomial time.

Epsilon-safe planning

We introduce an approach to high-level conditional planning we call e-safe planning. This probabilistic approach commits us to planning to meet some specified goal witch a probability of success of at least 1 - e for some user-supplied e. We describe several algorithms for e-safe planning based on conditional planners. The two conditional planners we discuss are Peot and Smiths nonlinear conditional planner, CNLP, and our own linear conditional planner, PLINTH. We present a straightforward extension to conditional planners for which computing the necessary probabilities is simple, employing a commonly-made but perhaps overly-strong independence assumption. We also discuss a second approach to e-safe planning which relaxes this independence assumption, involving the incremental construction of a probability dependence model in conjunction with the construction of the plan graph.

Temporal reasoning for planning and scheduling

We briefly describe the current implementation of the Time Map Manager (TMM), followed by a description of the systems suitability for and application to planning and scheduling tasks.

A constraint-based scheduler for batch manufacturing

Batch manufacturing poses unique challenges to schedulers. The manufacturing processes are unpredictable, the environment is dynamic, and the required task and resource models are complicated. The Honeywell batch scheduler uses constraint envelope scheduling to address these needs, offering support for both schedule modifications and rescheduling. The authors are reimplementing the scheduler as production quality software. They describe the process of building the scheduler and lessons learned during this process.

Representing uncertainty in simple planners

In this paper, we present an analysis of planning with uncertain information regarding both the state of the world and the effects of actions using a Strips- or (propositional) ADL-style representation [4, 17]. We provide formal definitions of plans under incomplete information and conditional plans, and describe PLINTH, a conditional linear planner based on these definitions. We also clarify the definition of the term “conditional action,” which has been variously used to denote actions with context-dependent effects and actions with uncertain outcomes. We show that the latter can, in theory, be viewed as a special case of the former but that to do so requires one to sacrifice the simple, single-model representation for one which can distinguish between a proposition and beliefs about that proposition.

A New Method for the Global Solution of Large Systems of Continuous Constraints

Scheduling of refineries is a hard hybrid problem. Application of the Constraint Envelope Scheduling (CES) approach required development of the Gradient Constraint Equation Subdivision (GCES) algorithm, a novel global feasibility solver for the large system of quadratic constraints that arise as subproblems. We describe the implemented solver and its integration into the scheduling system. We include discussion of pragmatic design tradeoffs critically important to achieving reasonable performance.

Planning applications in image analysis

Abstract We describe two interim results from an ongoing effort to automate the acquisition, analysis, archiving, and distribution of satellite Earth science data. Both results are applications of artificial intelligence planning research to the automatic generation of processing steps for image analysis tasks. First, we have constructed a linear conditional planner (PLINTH), and used it to generate conditional plans for image-processing. Second, we have extended an existing hierarchical planning system to make use of durations, resources, and deadlines, thus supporting the automatic generation of processing steps in time- and resource-constrained environments.

Frankencode: Creating Diverse Programs Using Code Clones

In this paper, we present an approach to detecting novel cyber attacks though a form of program diversification, similar to the use of n-version programming for fault tolerant systems. Building on extensive previous and ongoing work by others on the use of code clones in a wide variety of areas, our Functionally Equivalent Variants using Information Synchronization (FEVIS) system automatically generates program variants to berun in parallel, seeking to detect attacks through divergence in behavior. Unlike approaches to diversification that only change program memory layout and behavior, FEVIS can detect attacks exploiting vulnerabilities in execution timing, string processing, and other logic errors. We are in the early stages of research and development for this approach, but have made sufficient progress to provide a proof of concept and some lessons learned. In this paper we describe FEVIS and its application to diversifying an open-source webserver, with results on several different example classes of attack which FEVIS will detect.

A method for global optimization of large systems of quadratic constraints

In previous work, we have presented a novel global feasibility solver for the large system of quadratic constraints that arise as subproblems in the solving of hard hybrid problems, such as the scheduling of refineries. In this paper we present the Gradient Optimal Constraint Equation Subdivision (GOCES) algorithm, which incorporates a standard NLP solver and the global feasibility solver to find and establish global optimums for systems of quadratic equations, and present benchmarks.