SEARCH · Search NASA
Search indexed NASA NTRS and DOE OSTI research on propulsion, heat transfer, battery materials and energy systems. Follow report and document links to the original sources.
Quote a phrase for an exact phrase match. Source license links do not imply unrestricted reuse.
In this paper, we we analyze a number of tie-breaking staragegies for GSAT and evaluate the strategies for GSAT and evaluate the strategies empirically usuing randomly generated 3-SAT instances from a hard distribution of random instances.
An algorithm for using electric currents to control pendular oscillations induced by various perturbing forces on the Skyhook wire is considered. Transverse and vertical forces on the tether; tether instability modes and causes during retrieval by space shuttle; simple and spherical pendulum motion and vector damping; and current generation and control are discussed. A computer program for numerical integration of the in-plane and out-of-plane displacements of the tether vs time was developed for heuristic study. Some techniques for controlling instabilities during payload retrieval and methods for employing the tether for launching satellites from the space shuttle are considered. Derivations and analyses of a general nature used in all of the areas studied are included.
An algorithm for automating correspondence detection between point clouds composed of multibeam sonar data is presented. This allows accurate initialization for point cloud alignment techniques even in cases where accurate inertial navigation is not available, such as iceberg profiling or vehicles with low-grade inertial navigation systems. Techniques from computer vision literature are used to extract, label, and match keypoints between "pseudo-images" generated from these point clouds. Image matches are refined using RANSAC and information about the vehicle trajectory. The resulting correspondences can be used to initialize an iterative closest point (ICP) registration algorithm to estimate accumulated navigation error and aid in the creation of accurate, self-consistent maps. The results presented use multibeam sonar data obtained from multiple overlapping passes of an underwater canyon in Monterey Bay, California. Using strict matching criteria, the method detects 23 between-swath correspondence events in a set of 155 pseudo-images with zero false positives. Using less conservative matching criteria doubles the number of matches but introduces several false positive matches as well. Heuristics based on known vehicle trajectory information are used to eliminate these.
This paper presents an algorithm (WHAMII) designed to solve the Artificial Intelligence Design Challenge at the 1987 AIAA Guidance, Navigation and Control Conference. The problem under consideration is a stochastic generalization of the traveling salesman problem in which travel costs can incur a penalty with a given probability. The variability in travel costs leads to a probability constraint with respect to violating the budget allocation. Given the small size of the problem (eleven cities), an approach is considered that combines partial tour enumeration with a heuristic city insertion procedure. For computational efficiency during both the enumeration and insertion procedures, precalculated binomial probabilities are used to determine an upper bound on the actual probability of violating the budget constraint for each tour. The actual probability is calculated for the final best tour, and additional insertions are attempted until the actual probability exceeds the bound.
This paper is a new step in the development and application of the Linguistic Geometry. This formal theory is intended to discover the inner properties of human expert heuristics, which have been successful in a certain class of complex control systems, and apply them to different systems. In this paper we investigate heuristics extracted in the form of hierarchical networks of planning paths of autonomous agents. Employing Linguistic Geometry tools the dynamic hierarchy of networks is represented as a hierarchy of formal attribute languages. The main ideas of this methodology are shown in this paper on the new pilot example of the solution of the extremely complex 3D optimization problem of strategic planning for the space combat of autonomous vehicles. This example demonstrates deep and highly selective search in comparison with conventional search algorithms.
In this chapter an analysis of the behavior of an arbitrary (perhaps massive) collective of computational processes in terms of an associated "world" utility function is presented We concentrate on the situation where each process in the collective can be viewed as though it were striving to maximize its own private utility function. For such situations the central design issue is how to initialize/update the collective's structure, and in particular the private utility functions, so as to induce the overall collective to behave in a way that has large values of the world utility. Traditional "team game" approaches to this problem simply set each private utility function equal to the world utility function. The "Collective Intelligence" (COIN) framework is a semi-formal set of heuristics that recently have been used to construct private utility. functions that in many experiments have resulted in world utility values up to orders of magnitude superior to that ensuing from use of the team game utility. In this paper we introduce a formal mathematics for analyzing and designing collectives. We also use this mathematics to suggest new private utilities that should outperform the COIN heuristics in certain kinds of domains. In accompanying work we use that mathematics to explain previous experimental results concerning the superiority of COIN heuristics. In that accompanying work we also use the mathematics to make numerical predictions, some of which we then test. In this way these two papers establish the study of collectives as a proper science, involving theory, explanation of old experiments, prediction concerning new experiments, and engineering insights.
Grid generation plays an integral part in the solution of computational fluid dynamics problems for aerodynamics applications. A major difficulty with standard structured grid generation, which produces quadrilateral (or hexahedral) elements with implicit connectivity, has been the requirement for a great deal of human intervention in developing grids around complex configurations. This has led to investigations into unstructured grids with explicit connectivities, which are primarily composed of triangular (or tetrahedral) elements, although other subdivisions of convex cells may be used. The existence of large gradients in the solution of aerodynamic problems may be exploited to reduce the computational effort by using high aspect ratio elements in high gradient regions. However, the heuristic approaches currently in use do not adequately address this need for high aspect ratio unstructured grids. High aspect ratio triangulations very often produce the large angles that are to be avoided. Point generation techniques based on contour or front generation are judged to be the most promising in terms of being able to handle complicated multiple body objects, with this technique lending itself well to adaptivity. The eventual goal encompasses several phases: first, a partitioning phase, in which the Voronoi diagram of a set of points and line segments (the input set) will be generated to partition the input domain; second, a contour generation phase in which body-conforming contours are used to subdivide the partition further as well as introduce the foundation for aspect ratio control, and; third, a Steiner triangulation phase in which points are added to the partition to enable triangulation while controlling angle bounds and aspect ratio. This provides a combination of the advancing front/contour techniques and refinement. By using a front, aspect ratio can be better controlled. By using refinement, bounds on angles can be maintained, while attempting to minimize the number of Steiner points.
We present a systematic design method executed under resource constraints for automating the design of artificial neural networks using the back error propagation algorithm. Our system aims at finding the best possible configuration for solving the given application with proper tradeoff between the training time and the network complexity. The design of such a system is hampered by three related problems. First, there are infinitely many possible network configurations, each may take an exceedingly long time to train; hence, it is impossible to enumerate and train all of them to completion within fixed time, space, and resource constraints. Second, expert knowledge on predicting good network configurations is heuristic in nature and is application dependent, rendering it difficult to characterize fully in the design process. A learning procedure that refines this knowledge based on examples on training neural networks for various applications is, therefore, essential. Third, the objective of the network to be designed is ill-defined, as it is based on a subjective tradeoff between the training time and the network cost. A design process that proposes alternate configurations under different cost-performance tradeoff is important. We have developed a Design System which schedules the available time, divided into quanta, for testing alternative network configurations. Its goal is to select/generate and test alternative network configurations in each quantum, and find the best network when time is expended. Since time is limited, a dynamic schedule that determines the network configuration to be tested in each quantum is developed. The schedule is based on relative comparison of predicted training times of alternative network configurations using comparator network paradigm. The comparator network has been trained to compare training times for a large variety of traces of TSSE-versus-time collected during back-propagation learning of various applications.
We develop a formal theory, the so-called Linguistic Geometry, in order to discover the inner properties of human expert heuristics, which were successful in a certain class of complex control systems, and apply them to different systems. This research relies on the formalization of search heuristics of high-skilled human experts which allow for the decomposition of complex system into the hierarchy of subsystems, and thus solve intractable problems reducing the search. The hierarchy of subsystems is represented as a hierarchy of formal attribute languages. This paper includes a formal survey of the Linguistic Geometry, and new example of a solution of optimization problem for the space robotic vehicles. This example includes actual generation of the hierarchy of languages, some details of trajectory generation and demonstrates the drastic reduction of search in comparison with conventional search algorithms.
Many data-driven planning methods are trained on data generated by planners. It is well known that many statistical learning methods are sensitive to sampling bias, and yet there has been little or no attention to planning as a sampling method and its role in introducing sampling bias into planner-generated training data. Recently, it has been demonstrated that A**,* in the presence of problems with variable heuristic error, prefers some solutions over other equally cost-optimal solutions. But, as we discuss in this paper, mitigation may not be as simple as resolving arbitrary tie-breaking by sampling from ties uniformly at random. In this paper, we formalize an intuition of planning bias. We focus on problems which output a single solution. Diverse planning only complicates the problem by generalizing it to bias in the set of sets; we show how it is subject to bias in the single solution. We make some useful observations about deterministic algorithms in contrast to non-deterministic algorithms. We explain how information entropy may be a good way to measure planning bias, and discuss some issues in evaluating practical approaches to measurement. We address the intuition that uniform random tiebreaking should mitigate bias; and sketch a novel approach to constructing an appropriate random distribution for duplicate detection during forward search for unbiased A*. Finally, we suggest directions for future work.
We have developed a novel technique to incorporate uncertainty modeling within an evolutionary algorithm approach to multi-objective scheduling, with the goal of identifying a Pareto frontier (tradeoff curve) that recognizes the likelihood of events that can impact the schedule outcome. Our approach is particularly applicable to the generation of multiobjective optimized robust schedules, where objectives are assigned a service level, for example that we require an objective value to be greater than or equal to X with Y% confidence. We have demonstrated that such an approach can, for example, minimize scheduling on less reliable resources, based solely on a resource reliability model and not on any ad hoc heuristics. We have also investigated an alternative method of optimizing for robustness, in which we add to the set of objectives a failure risk objective to minimize. We compare the advantages and disadvantages of these two approaches. Future plans for further developing this technology include its application to space-based observatory scheduling problems.
New heuristics are integrated with four-dimensional guidance techniques developed by NASA in order to automate the prediction and resolution of airspace violations, and three rules are derived to reduce the amount of computation time for conflict detection. Controller experience has been used to establish techniques for resolving potential conflicts including speed control, altitude maneuvers, and horizontal flight maneuvers. Current aircraft states, flight plans, and extrapolation of aircraft trajectories are used to synthesize four-dimensional trajectories, and the state of the air traffic and the interaction between aircraft are projected into the future using the four-dimensional algorithms. The situation is immediately updated to incorporate any new information. Potential conflicts among multiple aircraft are resolved by fast sequential resolution of potential conflicts between aircraft pairs, one at a time.
Spaceborne computing systems must provide reliable, continuous operation for extended periods. Due to weight, power, and volume constraints, these systems must manage resources very effectively. A fault diagnosis algorithm is described which enables fast and flexible diagnoses in the dynamic distributed computing environments planned for future space missions. The algorithm uses a knowledge base that is easily changed and updated to reflect current system status. Augmented fault trees represented in an object-oriented form provide deep system knowledge that is easy to access and revise as a system changes. Given such a fault tree, a set of failure events that have occurred, and a set of failure events that have not occurred, this diagnosis system uses forward and backward chaining to propagate causal and temporal information about other failure events in the system being diagnosed. Once the system has established temporal and causal constraints, it reasons backward from heuristically selected failure events to find a set of basic failure events which are a likely cause of the occurrence of the top failure event in the fault tree. The diagnosis system has been implemented in common LISP using Flavors.
Calibration of computational models in the presence of uncertainty is often cast as a Bayesian inference problem and solved via sampling methods, e.g., Markov chain Monte Carlo. When the computational model is expensive, this task becomes intractable due to the large number of samples required to accurately estimate the posterior distribution of the calibration parameters. A popular solution to this problem is to use machine learning to develop a faster-to-evaluate, lower-fidelity substitute for the original model to serve as a surrogate while solving the inference problem. Although considered an offline cost, generating training data to construct this surrogate model can still be an expensive task in practice. An active learning algorithm is presented that focuses training on improving surrogate accuracy specifically in and around the bulk of the posterior distribution, as this is where the model is exercised during calibration. Candidate samples are drawn from families of distributions related to an approximation of the posterior. The sample maximizing predictive variance is then selected for evaluation by the original computational model, yielding a label for the training point. Iterating this approach increases efficiency relative to space filling designs (e.g., Latin hypercube sampling) by avoiding low probability points. Practical considerations are discussed, including the benefits of using a sequential Monte Carlo sampling approach, convergence heuristics, and the importance of both exploration and exploitation given that the true posterior is unknown a priori.
The demands placed on the mass storage systems at various federal agencies and national laboratories are continuously increasing in intensity. This forces system managers to constantly monitor the system, evaluate the demand placed on it, and tune it appropriately using either heuristics based on experience or analytic models. Performance models require an accurate workload characterization. This can be a laborious and time consuming process. It became evident from our experience that a tool is necessary to automate the workload characterization process. This paper presents the design and discusses the implementation of a tool for workload characterization of mass storage systems. The main features of the tool discussed here are: (1)Automatic support for peak-period determination. Histograms of system activity are generated and presented to the user for peak-period determination; (2) Automatic clustering analysis. The data collected from the mass storage system logs is clustered using clustering algorithms and tightness measures to limit the number of generated clusters; (3) Reporting of varied file statistics. The tool computes several statistics on file sizes such as average, standard deviation, minimum, maximum, frequency, as well as average transfer time. These statistics are given on a per cluster basis; (4) Portability. The tool can easily be used to characterize the workload in mass storage systems of different vendors. The user needs to specify through a simple log description language how the a specific log should be interpreted. The rest of this paper is organized as follows. Section two presents basic concepts in workload characterization as they apply to mass storage systems. Section three describes clustering algorithms and tightness measures. The following section presents the architecture of the tool. Section five presents some results of workload characterization using the tool.Finally, section six presents some concluding remarks.
Implementation of computer programs based on multivariate statistical algorithms makes possible obtaining reliable information from long data vectors that contain large amounts of extraneous information, for example, noise and/or analytes that we do not wish to control. Three examples are described. Each of these applications requires the use of techniques characteristic of modern analytical chemistry. The first example, using a quantitative or analytical model, describes the determination of the acid dissociation constant for 2,2'-pyridyl thiophene using archived data. The second example describes an investigation to determine the active biocidal species of iodine in aqueous solutions. The third example is taken from a research program directed toward advanced fiber-optic chemical sensors. The second and third examples require heuristic or empirical models.
At the Honeywell Technology Center (HTC), we have been working on a scheduling problem related to commercial avionics. This application is large, complex, and hard to solve. To be a little more concrete: 'large' means almost 20,000 activities, 'complex' means several activity types, periodic behavior, and assorted types of temporal constraints, and 'hard to solve' means that we have been unable to eliminate backtracking through the use of search heuristics. At this point, we can generate solutions, where solutions exist, or report failure and sometimes why the system failed. To the best of our knowledge, this is among the largest and most complex scheduling problems to have been solved as a constraint satisfaction problem, at least that has appeared in the published literature. This abstract is a preliminary report on what we have done and how. In the next section, we present our approach to treating scheduling as a constraint satisfaction problem. The following sections present the application in more detail and describe how we solve scheduling problems in the application domain. The implemented system makes use of Ginsberg's Dynamic Backtracking algorithm, with some minor extensions to improve its utility for scheduling. We describe those extensions and the performance of the resulting system. The paper concludes with some general remarks, open questions and plans for future work.
Sensor based robot motion planning research has primarily focused on mobile robots. Consider, however, the case of a robot manipulator expected to operate autonomously in a dynamic environment where unexpected collisions can occur with many parts of the robot. Only a sensor based system capable of generating collision free paths would be acceptable in such situations. Recently, work in this area has been reported in which a deterministic solution for 2DOF systems has been generated. The arm was sensitized with 'skin' of infra-red sensors. We have proposed a heuristic (potential field based) methodology for redundant robots with large DOF's. The key concepts are solving the path planning problem by cooperating global and local planning modules, the use of complete information from the sensors and partial (but appropriate) information from a world model, representation of objects with hyper-ellipsoids in the world model, and the use of variational planning. We intend to sensitize the robot arm with a 'skin' of capacitive proximity sensors. These sensors were developed at NASA, and are exceptionally suited for the space application. In the first part of the report, we discuss the development and modeling of the capacitive proximity sensor. In the second part we discuss the motion planning algorithm.