Nilufer Onder

Contingency selection in plan generation

Planning under uncertainty is a ubiquitous subarea of the recent planning literature. Systems that plan under uncertainty provide mechanisms for generating contingency plans: conditional planners can generate branching plans by allowing for actions with multiple possible outcomes (thus possible failures) and sensing actions that let the agent determine the current state. A key question in conditional planning is: how many, and which of the possible execution failures should be planned for by including conditional branches? This thesis addresses the question of selecting contingencies to repair with conditional branches before plan execution starts. A closely related issue is the identification of the types of repairs that can be performed once a contingency is selected. In the thesis, recent work on conditional and/or probabilistic planning is synthesized into a unifying algorithm that incorporates and clarifies the main techniques that have been developed. The main feature of the algorithm is the decoupling of the search-control strategy from the underlying plan-refinement process by including two direct methods for repairing possible failure points among the plan-refinement operations. The first method is corrective repair: the process of using steps to try to correct the situation after a failure. The second method is preventive repair: the process of using steps to try to reduce the chance of failure. The algorithm is fully implemented in a planning system called Mahinur and the system is used as a framework for principled prioritization of repairs for failure points. This prioritization is achieved by using a decision theoretic account of the expected value of performing the repairs considering the probability of failure and the impact of the failure on the top-level goals assigned to the agent. The benefits of contingency selection axe shown experimentally in synthetic domains and a realistic domain involving oil spill response plans. The effectiveness of additional conditional, probabilistic planning heuristics are analyzed and demonstrated empirically.

Engineering a conformant probabilistic planner

We present a partial-order, conformant, probabilistic planner, Probapop which competed in the blind track of the Probabilistic Planning Competition in IPC-4. We explain how we adapt distance based heuristics for use with probabilistic domains. Probapop also incorporates heuristics based on probability of success. We explain the successes and diefficulties encountered during the design and implementation of Probapop.

Using situational simulations to collect and analyze dynamic construction management decision-making data

In this paper we lay the foundations for studying decision-making in complex dynamic construction management scenarios using situational simulations as experimental testbeds. We draw on research conducted in dynamic decision making, construction data-mining and situational simulations to develop methods to study human decision-making data collected in ICDMA - a situational simulation of a real four story steel frame office building construction project. Specifically, we address challenges in the collection, organization and analysis of human subject data. We define a discipline driving the collection of human decision-making data, establish a semantics to organize the data and a simple mathematical syntax to represent it. We also present an analysis of preliminary experimental work and show that our method can be used to analyze patterns in complex construction decision-making. Finally, we present an agenda of research in construction decision-making using situational simulations that can be conducted using our proposed methods.

Expecting the unexpected: representing, reasoning about, and assessing construction project contingencies

Planning, scheduling and effective management of contingencies are crucial for the successful management of construction projects. In this paper we explore a mathematical representation of construction processes that can be used to infer alternative futures of a project as it unfolds. The representation has its foundations in temporal constraint networks. We present algorithms that can traverse the network in time, reason about the constraints driving a construction project, and present the combinatorial possibilities of futures that can emerge from one or more constraint violations during project implementation. The algorithms will aid construction managers to anticipate and react to crisis scenarios as they evolve in time. Our broader goal is to use the contingency information and the user responses to reveal the cognitive strategies used by humans to manage complex crisis scenarios.

Strategy optimization and generation for construction project management using an interactive simulation

Construction activities are exposed to unpredictable external events and internal dynamic feedbacks of constraints which deviate projects from as-planned duration and costs. In practice, various decisions have been used to minimize the impacts of risks. Learning from experiences is valuable which requires historical data collection and analysis. To avoid the high cost in direct data collection and difficulties in studying single decision impacts, we present an alternative to study and optimize decision strategies. Interactive Construction Decision Making Aid (ICDMA) is an interactive simulator which allows users to implement different decision strategies on the defined projects. All the project information and decision data in the simulation are recorded electronically. We started with five candidate strategies and analyzed data for general patterns. New hybrid strategies were generated based on the data analysis. Reimplementation of new strategies showed improvement in cost and duration management, validating the feasibility of strategy optimization through interactive simulation.

A multipurpose simulation platform for decision-making in construction management

Research in general purpose and special purpose simulation platforms typically treat model development, experimentation, validation and deployment of simulations as distinct phases. Direct involvement of decision-makers is usually limited to the validation phase, even though their participation significantly improves the effectiveness and applicability of models. Unfortunately, the complexity and sophistication involved in the model development and experimentation phases deters their participation. This also makes the validation problem particularly challenging, and hinders the credibility and successful deployment of such models. In addressing this problem, we introduce an interactive simulation platform called Interactive Construction Decision Making Aid (ICDMA), that integrates decision-maker participation into the model development, simulation deployment, experimentation and validation phases. Effectively it separates the complexities of programming the model and the model development process, thus encouraging the participation of domain experts. We illustrate the usefulness of this platform.

Studying Dynamic Decision-Making in Construction Management Using Adaptive Interactive Simulations

Construction projects consist of multiple components such as on-site equipment resources, site layout, labor productivity, unexpected external events, and human decisions regarding resource allocation and activity rescheduling. Stochastic spatial and temporal interactions between those components result in dynamic complexity of construction projects. Decisions made in such dynamic task environments must consider both immediate impacts and long-term dynamic feedback. The objective of this research is to use the adaptive interactive simulation framework as an experimental test bed to study dynamics of decisions in construction scenarios. A simulation of a four story structural steel framed office building project was implemented within Interactive Construction Decision Making Aid (ICDMA) under five different decision strategies. Simulated construction performance data at each decision control point was collected and analyzed to compare the project dynamics resulting from five strategies, and those strategies were assessed as well. This research provides a perspective to study dynamics of decision making through interactive simulation while furthering the knowledge of effective decision making and the development of dynamic decision support systems in construction management.

An experimental framework to analyze alternative decision-making strategies using situational simulations in construction management

Situational simulations are dynamic, interactive, context-sensitive, adaptive environments. They further construction research by providing an interactive simulation platform that can be used to explore what-if construction scenarios and to estimate risks and contingencies. This paper extends current research to study the evolution of dynamic uncertainty in construction management projects using situational simulations as experimental testbeds. An experimental framework is proposed to explore alternative outcomes of a particular decision strategy, and also investigate the impact of alternative decision strategies under similar project scenarios. A set of preliminary experiments were conducted to illustrate the proposed framework. The significance of this research is in enhancing and informing the deliberative process during the planning and pre-planning stages of a construction project and supporting the preparation of contingency plans of action in anticipation of varying levels of project risk and uncertainty.

mPart: miss-ratio curve guided partitioning in key-value stores

Web applications employ key-value stores to cache the data that is most commonly accessed. The cache improves an web applications performance by serving its requests from memory, avoiding fetching them from the backend database. Since the memory space is limited, maximizing the memory utilization is a key to delivering the best performance possible. This has lead to the use of multi-tenant systems, allowing applications to share cache space. In addition, application data access patterns change over time, so the system should be adaptive in its memory allocation. In this work, we address both multi-tenancy (where a single cache is used for multiple applications) and dynamic workloads (changing access patterns) using a model that relates the cache size to the application miss ratio, known as a miss ratio curve. Intuitively, the larger the cache, the less likely the system will need to fetch the data from the database. Our efficient, online construction of the miss ratio curve allows us to determine a near optimal memory allocation given the available system memory, while adapting to changing data access patterns. We show that our model outperforms an existing state-of-the-art sharing model, Memshare, in terms of overall cache hit ratio and does so at a lower time cost. We show that for a typical system, overall hit ratio is consistently 1 percentage point greater and 99.9th percentile latency is reduced by as much as 2.9% under standard web application workloads containing millions of requests.

Structured-Chromosome Evolutionary Algorithms for Variable-Size Autonomous Interplanetary Trajectory Planning Optimization

In interplanetary trajectory optimization, events such as planetary gravitational-assist maneuvers (swingbys) and deep-space maneuvers can be added/removed from the trajectory plan to reduce the cost or the flight time. This renders the number of design variables in the optimization problem variable. Global optimization methods that optimize this type of multimodal objective function can only handle problems with a fixed number of design variables. This paper presents the structured-chromosome evolutionary algorithm framework that is developed to handle variable-size design space optimization problems. In this framework, a solution (chromosome) is represented by a hierarchical data structure where the genes in the chromosome are classified as dependent and nondependent genes. This structure provides the capability to apply genetic operations between solutions of different lengths, and thus to automatically determine the number of swingbys, the planets to swingby, launch and arrival dates, and the number o...