Search NASA⌕ Search

SEARCH · Search NASA

Results for “deep generative models”

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.

At least 1,351 records · Page 75

Operator learning for predicting multiscale bubble growth dynamics

We report simulating and predicting multiscale problems that couple multiple physics and dynamics across many orders of spatiotemporal scales is a great challenge that has not been investigated systematically by deep neural networks (DNNs). Herein, we develop a framework based on operator regression, the so-called deep operator network (DeepONet), with the long-term objective to simplify multiscale modeling by avoiding the fragile and time-consuming “hand-shaking” interface algorithms for stitching together heterogeneous descriptions of multiscale phenomena. To this end, as a first step, we investigate if a DeepONet can learn the dynamics of different scale regimes, one at the deterministic macroscale and the other at the stochastic microscale regime with inherent thermal fluctuations. Specifically, we test the effectiveness and accuracy of the DeepONet in predicting multirate bubble growth dynamics, which is described by a Rayleigh–Plesset (R–P) equation at the macroscale and modeled as a stochastic nucleation and cavitation process at the microscale by dissipative particle dynamics (DPD). First, we generate data using the R–P equation for multirate bubble growth dynamics caused by randomly time-varying liquid pressures drawn from Gaussian random fields (GRFs). Our results show that properly trained DeepONets can accurately predict the macroscale bubble growth dynamics and can outperform long short-term memory networks. We also demonstrate that the DeepONet can extrapolate accurately outside the input distribution using only very few new measurements. Subsequently, we train the DeepONet with DPD data corresponding to stochastic bubble growth dynamics. Although the DPD data are noisy and we only collect sparse data points on the trajectories, the trained DeepONet model is able to predict accurately the mean bubble dynamics for time-varying GRF pressures. Taken together, our findings demonstrate that DeepONets can be employed to unify the macroscale and microscale models of the multirate bubble growth problem, hence providing new insight into the role of operator regression via DNNs in tackling realistic multiscale problems and in simplifying modeling with heterogeneous descriptions.

37 INORGANIC, ORGANIC, PHYSICAL, AND ANALYTICAL CH↗

TYPE Iax SUPERNOVAE: A NEW CLASS OF STELLAR EXPLOSION

We describe observed properties of the Type Iax class of supernovae (SNe Iax), consisting of SNe observationally similar to its prototypical member, SN 2002cx. The class currently has 25 members, and we present optical photometry and/or optical spectroscopy for most of them. SNe Iax are spectroscopically similar to SNe Ia, but have lower maximum-light velocities (2000 {approx}< |v| {approx}< 8000 km s{sup -1}), typically lower peak magnitudes (-14.2 {>=} M{sub V,{sub peak}} {approx}> -18.9 mag), and most have hot photospheres. Relative to SNe Ia, SNe Iax have low luminosities for their light-curve shape. There is a correlation between luminosity and light-curve shape, similar to that of SNe Ia, but offset from that of SNe Ia and with larger scatter. Despite a host-galaxy morphology distribution that is highly skewed to late-type galaxies without any SNe Iax discovered in elliptical galaxies, there are several indications that the progenitor stars are white dwarfs (WDs): evidence of C/O burning in their maximum-light spectra, low (typically {approx}0.5 M{sub Sun }) ejecta masses, strong Fe lines in their late-time spectra, a lack of X-ray detections, and deep limits on massive stars and star formation at the SN sites. However, two SNe Iax show strong He lines in their spectra. The progenitor system and explosion model that best fits all of the data is a binary system of a C/O WD that accretes matter from a He star and has a deflagration. At least some of the time, this explosion will not disrupt the WD. The small number of SNe in this class prohibit a detailed analysis of the homogeneity and heterogeneity of the entire class. We estimate that in a given volume there are 31{sup +17}{sub -13} SNe Iax for every 100 SNe Ia, and for every 1 M{sub Sun} of iron generated by SNe Ia at z = 0, SNe Iax generate {approx}0.036 M{sub Sun }. Being the largest class of peculiar SNe, thousands of SNe Iax will be discovered by LSST. Future detailed observations of SNe Iax should further our understanding of both their progenitor systems and explosions as well as those of SNe Ia.

79 ASTRONOMY AND ASTROPHYSICS↗

Optimizing and Quantifying CO 2 Storage Resource in Saline Formations and Hydrocarbon Reservoirs

In an effort to reduce carbon dioxide (CO 2 ) emissions from large stationary sources, carbon capture and storage (CCS) is being investigated as one approach. This work assesses CO 2 storage resource estimation methods for deep saline formations (DSFs) and hydrocarbon reservoirs undergoing CO 2 enhanced oil recovery (EOR). Project activities were conducted using geologic modeling and simulation to investigate CO 2 storage efficiency. CO 2 storage rates and efficiencies in DSFs classified by interpreted depositional environment were evaluated at the regional scale over a 100-year time frame. A focus was placed on developing results applicable to future widespread commercial-scale CO 2 storage operations in which an array of injection wells may be used to optimize storage in saline formations. The results of this work suggest future investigations of prospective storage resource in closed or semiclosed formations need not have a detailed understanding of the depositional environment of the reservoir to generate meaningful estimates. However, the results of this work also illustrate the relative importance of depositional environment, formation depth, structural geometry, and boundary conditions on the rate of CO 2 storage in these types of systems. CO 2 EOR occupies an important place in the realm of geologic storage of CO 2 , as it is likely to be the primary means of geologic CO 2 storage during the early stages of commercial implementation, given the lack of a national policy and the viability of the current business case. This work estimates CO 2 storage efficiency factors using a unique industry database of CO 2 EOR sites and 18 different reservoir simulation models capturing fluvial clastic and shallow shelf carbonate depositional environments for reservoir depths of 1219 and 2438 meters (4000 and 8000 feet) and 7.6-, 20-, and 64-meter (25-, 66,- and 209-foot) pay zones. The results of this work provide practical information that can be used to quantify CO 2 storage resource estimates in oil reservoirs during CO 2 EOR operations (as opposed to storage following depletion) and the uncertainty associated with those estimates.

54 ENVIRONMENTAL SCIENCES↗

A coupled dynamic-thermodynamic model of an ice-ocean system in the marginal ice zone

Thermodynamics are incorporated into a coupled ice-ocean model in order to investigate wind-driven ice-ocean processes in the marginal zone. Upswelling at the ice edge which is generated by the difference in the ice-air and air-water surface stresses is found to give rise to a strong entrainment by drawing the pycnocline closer to the surface. Entrainment is shown to be negligible outside the areas affected by the ice edge upswelling. If cooling at the top is included in the model, the heat and salt exchanges are further enhanced in the upswelling areas. It is noted that new ice formation occurs in the region not affected by ice edge upswelling, and it is suggested that the high-salinity mixed layer regions (with a scale of a few Rossby radii of deformation) will overturn due to cooling, possibly contributing to the formation of deep water.

Hakkinen, Sirpa↗

Mission Operations with an Autonomous Agent

The Remote Agent (RA) is an Artificial Intelligence (AI) system which automates some of the tasks normally reserved for human mission operators and performs these tasks autonomously on-board the spacecraft. These tasks include activity generation, sequencing, spacecraft analysis, and failure recovery. The RA will be demonstrated as a flight experiment on Deep Space One (DSI), the first deep space mission of the NASA's New Millennium Program (NMP). As we moved from prototyping into actual flight code development and teamed with ground operators, we made several major extensions to the RA architecture to address the broader operational context in which PA would be used. These extensions support ground operators and the RA sharing a long-range mission profile with facilities for asynchronous ground updates; support ground operators monitoring and commanding the spacecraft at multiple levels of detail simultaneously; and enable ground operators to provide additional knowledge to the RA, such as parameter updates, model updates, and diagnostic information, without interfering with the activities of the RA or leaving the system in an inconsistent state. The resulting architecture supports incremental autonomy, in which a basic agent can be delivered early and then used in an increasingly autonomous manner over the lifetime of the mission. It also supports variable autonomy, as it enables ground operators to benefit from autonomy when L'@ey want it, but does not inhibit them from obtaining a detailed understanding and exercising tighter control when necessary. These issues are critical to the successful development and operation of autonomous spacecraft.

Pell, Barney↗

A Thermally-Conductive Martian Core and Implications for Its Dynamo Cessation

Mars experienced a dynamo process that generated a global magnetic field ~4.3 (or earlier) to 3.6 billion years ago (Ga). The cessation of this dynamo strongly affected Mars’ history and is expected to be linked to thermochemical evolution of Mars’ iron-rich liquid core, which is strongly influenced by its thermal conductivity. Here, we directly measured thermal conductivities of solid iron-sulfur alloys to pressures relevant to the Martian core and temperatures to 1023 Kelvin. Our results show that a Martian core with 16 weight % sulfur has a thermal conductivity of ~19 to 32 Watt meter −1 Kelvin −1 from its top to the center, much higher than previously inferred from electrical resistivity measurements. Our modeled thermal conductivity profile throughout the Martian deep-mantle and core indicates a ~4- to 6-fold discontinuity across the core-mantle boundary. The core’s efficient cooling resulting from the depth-dependent, high conductivity diminishes thermal convection and forms thermal stratification, substantially contributing to cessation of Martian dynamo.

Wen-Pin Hsieh↗

Energetics at the Surface: Direct Optical Mapping of Core and Surface Electronic Structure in CdSe Quantum Dots Using Broadband Electronic Sum Frequency Generation Microspectroscopy

Understanding and controlling the electronic structure of nanomaterials is the key to tailoring their use in a wide range of practical applications. Despite this need, many important electronic states are invisible to conventional optical measurements and are typically identified indirectly based on their inferred impact on luminescence properties. This is especially common and important in the study of nanomaterial surfaces and their associated defects. Surface trap states play a crucial role in photophysical processes yet remain remarkably poorly understood. In this work we demonstrate for the first time that broadband electronic sum frequency generation (eSFG) microspectroscopy can directly map the optically bright and dark states of nanoparticles, including the elusive below gap states. This new approach is applied to model cadmium selenide (CdSe) quantum dots (QDs), where the energies of surface trap states have eluded direct optical characterization for decades. Our eSFG measurements show clear signatures of electronic transitions both above the band gap, which we assign to previously reported one- and two-photon transitions associated with the CdSe core, as well as broad spectral signatures below the band gap that are attributed to surface states. In addition to the core states, this analysis reveals two distinct distributions of below gap states, providing the first direct optical measurement of both shallow and deep surface states on this system. Lastly, chemical modification of the surfaces via oxidation results in the relative increase in the signals originating from the surface states. Overall, our eSFG experiments provide an avenue to directly map the entirety of the QD core and surface electronic structure, which is expected to open up opportunities to study how these materials are grown in situ and how surface states can be controlled to tune functionality.

77 NANOSCIENCE AND NANOTECHNOLOGY↗

A new technology for manufacturing scheduling derived from space system operations

A new technology for producing finite capacity schedules has been developed in response to complex requirements for operating space systems such as the Space Shuttle, the Space Station, and the Deep Space Network for telecommunications. This technology has proven its effectiveness in manufacturing environments where popular scheduling techniques associated with Materials Resources Planning (MRPII) and with factory simulation are not adequate for shop-floor work planning and control. The technology has three components. The first is a set of data structures that accommodate an extremely general description of a factory's resources, its manufacturing activities, and the constraints imposed by the environment. The second component is a language and set of software utilities that enable a rapid synthesis of functional capabilities. The third component is an algorithmic architecture called the Five Ruleset Model which accommodates the unique needs of each factory. Using the new technology, systems can model activities that generate, consume, and/or obligate resources. This allows work-in-process (WIP) to be generated and used; it permits constraints to be imposed or intermediate as well as finished goods inventories. It is also possible to match as closely as possible both the current factory state and future conditions such as promise dates. Schedule revisions can be accommodated without impacting the entire production schedule. Applications have been successful in both discrete and process manufacturing environments. The availability of a high-quality finite capacity production planning capability enhances the data management capabilities of MRP II systems. These schedules can be integrated with shop-floor data collection systems and accounting systems. Using the new technology, semi-custom systems can be developed at costs that are comparable to products that do not have equivalent functional capabilities and/or extensibility.

Hornstein, R. S.↗

Using High Frequency Passive Microwave, A-train, and TRMM Data to Evaluate Hydrometer Structure in the NASA GEOS-5 Data Assimilation System

Validating water vapor and prognostic condensate in global models remains a challenging research task. Model parameterizations are still subject to a large number of tunable parameters; furthermore, accurate and representative in situ observations are very sparse, and satellite observations historically have significant quantitative uncertainties. Progress on improving cloud / hydrometeor fields in models stands to benefit greatly from the growing inventory ofA-Train data sets. ill the present study we are using a variety of complementary satellite retrievals of hydrometeors to examine condensate produced by the emerging NASA Modem Era Retrospective Analysis for Research and Applications, MERRA, and its associated atmospheric general circulation model GEOS5. Cloud and precipitation are generated by both grid-scale prognostic equations and by the Relaxed Arakawa-Schubert (RAS) diagnostic convective parameterization. The high frequency channels (89 to 183.3 GHz) from AMSU-B and MRS on NOAA polar orbiting satellites are being used to evaluate the climatology and variability of precipitating ice from tropical convective anvils. Vertical hydrometeor structure from the Tropical Rainfall Measuring Mission (TRMM) and CloudSat radars are used to develop statistics on vertical hydrometeor structure in order to better interpret the extensive high frequency passive microwave climatology. Cloud liquid and ice water path data retrieved from the Moderate Resolution Imaging Spectroradiometer, MODIS, are used to investigate relationships between upper level cloudiness and tropical deep convective anvils. Together these data are used to evaluate cloud / ice water path, gross aspects of vertical hydrometeor structure, and the relationship between cloud extent and surface precipitation that the MERRA reanalysis must capture.

Robertson, Franklin↗

Predicting Two-Dimensional Airfoil Performance Using Graph Neural Networks

Computer simulations require the use of meshes to simulate geometries. These meshes capture important geometric features of the design and can be used in machine learning modeling. This report explores the use of graph neural networks (GNNs) to learn features from two-dimensional (2D) airfoil designs represented as a set of nodes connected using edges. This type of network is common in aerospace applications: most geometries are represented as a mesh in order to perform analysis. The objective of this work is to use GNNs to predict the performance of 2D airfoils generated using the program XFOIL. The predicted performance parameters include bulk quantities such as coefficients of lift (C L ), drag (C d , C dp ), moment (C m ), and node-specific quantities such as coefficient of pressure (C p ). In this report, a spline convolutional graph-based neural network is compared with deep learning neural networks to predict both bulk and node-specific quantities. The findings indicate the GNNs are able to predict bulk quantities quite well; however, when the number of outputs is increased, the deep neural network (DNN) proves to be better in its prediction capability. Two different normalization strategies were compared in the training of both GNNs and DNNs: minmax and standard deviation. In both types of networks, standard deviation scaling proved to be the best.

machine learning↗

Primordial black hole mass functions as a probe of cosmic origin

Here, we discuss a novel window to probe the origin of our universe via the mass functions of primordial black holes (PBHs). The mass functions of PBHs are simply estimated using the conventional Press-Schechter formalism for two paradigms of cosmic origin, including inflationary ΛCDM and bounce cosmology. The standard inflationary ΛCDM model cannot generate an appreciable number of massive PBHs; however, non-trivial inflation models with blue-tilted power spectra at small scales and matter bounce cosmology provide formation mechanisms for heavy PBHs, which in turn, may seed the observed supermassive black holes (SMBHs). By fitting the SMBH mass functions at high redshift (z ~ 6) derived from Sloan Digital Sky Survey (SDSS) and Canada-France High-z Quasar Survey (CFHQS) quasars, for two paradigms of cosmic origin, we derive constraints on the PBH density fraction $f$ PBH at z ~ 6 and the characteristic mass M $\ast$ , with the prior assumption that all SMBHs stem from PBHs. We demonstrate that this newly proposed procedure, relying on astronomical measurements that utilize deep-field surveys of SMBHs at high redshift, can be used to constrain models of cosmic origin. Additionally, although not the main focus of this paper, we evolve the mass function from z ~ 6 to z ~ 0 through an assumption of 3 × 10 8 -year Eddington’s accretion, and give a rough estimation of $f$ PBH at z ~ 0.

79 ASTRONOMY AND ASTROPHYSICS↗

Advances in Geologic Disposal System Modeling and Application to Crystalline Rock

The Used Fuel Disposition Campaign (UFDC) of the U.S. Department of Energy (DOE) Office of Nuclear Energy (NE), Office of Fuel Cycle Technology (OFCT) is conducting research and development (R&D) on geologic disposal of used nuclear fuel (UNF) and high-level nuclear waste (HLW). Two of the high priorities for UFDC disposal R&D are design concept development and disposal system modeling (DOE 2011). These priorities are directly addressed in the UFDC Generic Disposal Systems Analysis (GDSA) work package, which is charged with developing a disposal system modeling and analysis capability for evaluating disposal system performance for nuclear waste in geologic media (e.g., salt, granite, clay, and deep borehole disposal). This report describes specific GDSA activities in fiscal year 2016 (FY 2016) toward the development of the enhanced disposal system modeling and analysis capability for geologic disposal of nuclear waste. The GDSA framework employs the PFLOTRAN thermal-hydrologic-chemical multi-physics code and the Dakota uncertainty sampling and propagation code. Each code is designed for massively-parallel processing in a high-performance computing (HPC) environment. Multi-physics representations in PFLOTRAN are used to simulate various coupled processes including heat flow, fluid flow, waste dissolution, radionuclide release, radionuclide decay and ingrowth, precipitation and dissolution of secondary phases, and radionuclide transport through engineered barriers and natural geologic barriers to the biosphere. Dakota is used to generate sets of representative realizations and to analyze parameter sensitivity.

12 MANAGEMENT OF RADIOACTIVE AND NON-RADIOACTIVE W↗

High throughput, accurate gene annotation through AI and HPC-enabled structural analysis

With the advances in next generation sequencing technologies, the number of sequenced genomes is growing exponentially, resulting in a technology bottleneck for the translation of sequence information into usable hypotheses about the function of each gene. We have proposed leveraging our leadership high-performance computing (HPC) resources to help break this annotation bottleneck. Here we design an HPC-based framework to infer gene function from gene sequence by incorporating information about protein structure and interactions predicted by deep learning approaches. Accurate functional prediction and gene annotation using computational methods will facilitate breakthroughs in the genomic sciences essential to understanding and harnessing life processes in bacteria, fungi and plants. The development and applications of the state-of-the-art deep neural networks to protein structural modeling, interaction prediction, sequence comparison, and quality assessment of protein structural models will be made possible by leadership computational resources. These HPC-enabled bioinformatics and molecular modeling tools will provide powerful insights into molecular functions of genes.

59 BASIC BIOLOGICAL SCIENCES↗

MOS 2.0: Modeling the Next Revolutionary Mission Operations System

Designed and implemented in the 1980's, the Advanced Multi-Mission Operations System (AMMOS) was a breakthrough for deep-space NASA missions, enabling significant reductions in the cost and risk of implementing ground systems. By designing a framework for use across multiple missions and adaptability to specific mission needs, AMMOS developers created a set of applications that have operated dozens of deep-space robotic missions over the past 30 years. We seek to leverage advances in technology and practice of architecting and systems engineering, using model-based approaches to update the AMMOS. We therefore revisit fundamental aspects of the AMMOS, resulting in a major update to the Mission Operations System (MOS): MOS 2.0. This update will ensure that the MOS can support an increasing range of mission types, (such as orbiters, landers, rovers, penetrators and balloons), and that the operations systems for deep-space robotic missions can reap the benefits of an iterative multi-mission framework.12 This paper reports on the first phase of this major update. Here we describe the methods and formal semantics used to address MOS 2.0 architecture and some early results. Early benefits of this approach include improved stakeholder input and buy-in, the ability to articulate and focus effort on key, system-wide principles, and efficiency gains obtained by use of well-architected design patterns and the use of models to improve the quality of documentation and decrease the effort required to produce and maintain it. We find that such methods facilitate reasoning, simulation, analysis on the system design in terms of design impacts, generation of products (e.g., project-review and software-delivery products), and use of formal process descriptions to enable goal-based operations. This initial phase yields a forward-looking and principled MOS 2.0 architectural vision, which considers both the mission-specific context and long-term system sustainability.

ground support systems↗

Spectral Synthesis for Geostationary Satellite-to-Satellite Translation

Earth-observing satellites carrying multispectral sensors are widely used to monitor the physical and biological states of the atmosphere, land, and oceans. These satellites have different vantage points above the Earth and different spectral imaging bands resulting in inconsistent imagery from one to another. This presents challenges in building downstream applications. What if we could generate synthetic bands for existing satellites from the union of all domains? We tackle the problem of generating synthetic spectral imagery for multispectral sensors as an unsupervised image-to-image translation problem modeled with a variational autoencoder (VAE) and generative adversarial network (GAN) architecture. Our approach introduces a novel shared spectral reconstruction loss to constrain the high-dimensional feature space of multispectral images. Simulated experiments performed by dropping one or more spectral bands show that cross-domain reconstruction outperforms measurements obtained from a second vantage point. Our proposed approach enables the synchronization of multispectral data and provides a basis for more homogeneous remote sensing datasets.

Geostationary satellites↗

Linking basin-scale and pore-scale gas hydrate distribution patterns in diffusion-dominated marine hydrate systems: DIFFUSION-DRIVEN HYDRATE GROWTH IN SANDS

The goal of this study is to computationally determine the potential distribution patterns of diffusion-driven methane hydrate accumulations in coarse-grained marine sediments. Diffusion of dissolved methane in marine gas hydrate systems has been proposed as a potential transport mechanism through which large concentrations of hydrate can preferentially accumulate in coarse-grained sediments over geologic time. Using one-dimensional compositional reservoir simulations, we examine hydrate distribution patterns at the scale of individual sand layers (1 to 20 m thick) that are deposited between microbially active fine-grained material buried through the gas hydrate stability zone (GHSZ). We then extrapolate to two- dimensional and basin-scale three-dimensional simulations, where we model dipping sands and multilayered systems. We find that properties of a sand layer including pore size distribution, layer thickness, dip, and proximity to other layers in multilayered systems all exert control on diffusive methane fluxes toward and within a sand, which in turn impact the distribution of hydrate throughout a sand unit. In all of these simulations, we incorporate data on physical properties and sand layer geometries from the Terrebonne Basin gas hydrate system in the Gulf of Mexico. We demonstrate that diffusion can generate high hydrate saturations (upward of 90%) at the edges of thin sands at shallow depths within the GHSZ, but that it is ineffective at producing high hydrate saturations throughout thick (greater than 10 m) sands buried deep within the GHSZ. As a result, we find that hydrate in fine-grained material can preserve high hydrate saturations in nearby thin sands with burial.

03 NATURAL GAS↗

Hardware-accelerated inference for real-time gravitational-wave astronomy

The field of transient astronomy has seen a revolution with the first gravitational-wave detections and the arrival of multi-messenger observations they enabled. Transformed by the first detection of binary black hole and binary neutron star mergers, computational demands in gravitational-wave astronomy are expected to grow by at least a factor of two over the next five years as the global network of kilometer-scale interferometers are brought to design sensitivity. With the increase in detector sensitivity, real-time delivery of gravitational-wave alerts will become increasingly important as an enabler of multi-messenger followup. In this work, we report a novel implementation and deployment of deep learning inference for real-time gravitational-wave data denoising and astrophysical source identification. This is accomplished using a generic Inference-as-a-Service model that is capable of adapting to the future needs of gravitational-wave data analysis. Overall, our implementation allows seamless incorporation of hardware accelerators and also enables the use of commercial or private (dedicated) as-a-service computing. Based on our results, we propose a paradigm shift in low-latency and offline computing in gravitational-wave astronomy. Such a shift can address key challenges in peak-usage, scalability and reliability, and provide a data analysis platform particularly optimized for deep learning applications. The achieved sub-millisecond scale latency will also be relevant for any machine learning-based real-time control systems that may be invoked in the operation of near-future and next generation ground-based laser interferometers, as well as the front-end collection, distribution and processing of data from such instruments.

46 INSTRUMENTATION RELATED TO NUCLEAR SCIENCE AND ↗

A Generative Control Capability for a Model-based Executive

This paper describes Burton, a core element of a new generation of goal-directed model-based autonomous executives. This executive makes extensive use of component-based declarative models to analyze novel situations and generate novel control actions both at the goal and hardware levels. It uses an extremely efficient online propositional inference engine to efficiently determine likely states consistent with current observations and optimal target states that achieve high level goals. It incorporates a flexible generative control sequencing algorithm within the reactive loop to bridge the gap between current and target states. The system is able to detect and avoid damaging and irreversible situations, After every control action it uses its model and sensors to detect anomalous situations and immediately take corrective action. Efficiency is achieved through a series of model compilation and online policy construction methods, and by exploiting general conventions of hardware design that permit a divide and conquer approach to planning. The paper presents a formal characterization of Burton's capability, develops efficient algorithms, and reports on experience with the implementation in the domain of spacecraft autonomy. Burton is being incorporated as one of the key elements of the Remote Agent core autonomy architecture for Deep Space One, the first spacecraft for NASA's New Millenium program.

Williams, Brian C.↗