Search NASA⌕ Search

SEARCH · Search NASA

Results for “temporal computing”

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 361 records · Page 20
All sourcesNASA NTRSDOE OSTI
DOE OSTI2024

PMU Data Quality and Sensor Health Monitoring

Phasor Measurement Units (PMUs) play a critical role in the evolution of the electric power industry by providing high-precision, real-time monitoring of essential power system metrics. However, effectively detecting abnormalities and critical events from PMU data is a complex task, complicated by intricate temporal patterns, a scarcity of labeled data for training algo- rithms, and constraints on online computational power. In this study, we apply TranAD, an innovative algorithm that combines transformer architectures with the refinement of adversarial learning, to both synthetic and real-world PMU datasets for developing a data quality and sensor online health monitoring platform for utilities. Our findings reveal that TranAD not only provides efficient detection and localization but also enhances the detail with which abnormalities are detected, marking a a significant step forward in the field of clean data acquisition processes for power system monitoring

deep neural network, machine learning (ML)↗
NASA NTRS1992

Spatial and temporal adaptive procedures for the unsteady aerodynamic analysis of airfoils using unstructured meshes

An algorithm which combines spatial and temporal adaption for the time integration of the two dimensional Euler equations on unstructured meshes of triangles is presented. Spatial adaption involves mesh enrichment to add elements in high gradient regions of the flow and mesh coarsening to remove elements where they are no longer needed. Temporal adaption is a time accurate, local time stepping procedure which integrates the flow equations in each cell according to the local numerical stability constraint. The flow solver utilizes a four stage Runge-Kutta time integration scheme with an upwind flux-split spatial discretization. Results obtained using spatial and temporal adaption indicate that highly accurate solutions can be obtained with a significant savings of computing time over global time stepping.

Hooker, John R.↗
NASA NTRS1992

Spatial and temporal adaptive procedures for the unsteady aerodynamic analysis of airfoils using unstructured meshes

An algorithm which combines spatial and temporal adaption for the time integration of the two-dimensional Euler equations on unstructured meshes of triangles is presented. Spatial adaption involves mesh enrichment to add elements in high gradient regions of the flow and mesh coarsening to remove elements where they are no longer needed. Temporal adaption is a time accurate, local time stepping procedure which integrates the flow equations in each cell according to the local numerical stability constraint. The flow solver utilizes a four-stage Runge-Kutta time integration scheme with an upwind flux-split spatial discretization. Results obtained using spatial and temporal adaption indicate that highly accurate solutions can be obtained with a significant savings of computing time over global time stepping.

Hooker, John R.↗
NASA NTRS1985

Cloud motion derived winds: Their accuracy, coverage, and suggestions for future improvement

Nature has provided us with a natural and easily visible method of tracing atmospheric motion through the measurement of cloud velocities. This source of wind information has been available from geosynchronous satellites since the launch of the ATS-1 Spin Scan camera. This sensor provided adequate spatial and temporal resolution views of individual cloud systems that could represent the wind with useful accuracy. During the last decade, cloud motion derived winds have become part of the operational system as they are routinely provided to the National Meteorological Center as input to the global numerical models. The principal limitations of cloud motion winds are that they (1) can be measured within the limits that cloud motions can represent the wind only where trackable clouds exist, (2) require knowledge of the cloud height, (3) require high spatial and temporal resolution geosynchronous satellite systems with high attitude determination accuracy, and (4) need sophisticated interactive computer systems for the calculation of high resolution fields. These limitations are examined and suggestions are made for how this product could be improved. Also, uses of the data for mesoscale purposes are discussed.

Shenk, William E.↗
NASA NTRS1993

Computational fluid mechanics

Two papers are included in this progress report. In the first, the compressible Navier-Stokes equations have been used to compute leading edge receptivity of boundary layers over parabolic cylinders. Natural receptivity at the leading edge was simulated and Tollmien-Schlichting waves were observed to develop in response to an acoustic disturbance, applied through the farfield boundary conditions. To facilitate comparison with previous work, all computations were carried out at a free stream Mach number of 0.3. The spatial and temporal behavior of the flowfields are calculated through the use of finite volume algorithms and Runge-Kutta integration. The results are dominated by strong decay of the Tollmien-Schlichting wave due to the presence of the mean flow favorable pressure gradient. The effects of numerical dissipation, forcing frequency, and nose radius are studied. The Strouhal number is shown to have the greatest effect on the unsteady results. In the second paper, a transition model for low-speed flows, previously developed by Young et al., which incorporates first-mode (Tollmien-Schlichting) disturbance information from linear stability theory has been extended to high-speed flow by incorporating the effects of second mode disturbances. The transition model is incorporated into a Reynolds-averaged Navier-Stokes solver with a one-equation turbulence model. Results using a variable turbulent Prandtl number approach demonstrate that the current model accurately reproduces available experimental data for first and second-mode dominated transitional flows. The performance of the present model shows significant improvement over previous transition modeling attempts.

Hassan, H. A.↗
NASA NTRS1995

Special-Purpose High-Torque Permanent-Magnet Motors

Permanent-magnet brushless motors that must provide high commanded torques and satisfy unusual heat-removal requirement are developed. Intended for use as thrust-vector-control actuators in large rocket engines. Techniques and concepts used to design improved motors for special terrestrial applications. Conceptual motor design calls for use of rotor containing latest high-energy-product rare-earth permanent magnets so that motor produces required torque while drawing smallest possible currents from power supply. Torque generated by electromagnetic interaction between stator and permanent magnets in rotor when associated electronic circuits applied appropriately temporally and spatially phased currents to stator windings. Phase relationships needed to produce commanded torque computed in response to torque command and to electronically sensed angular position of rotor relative to stator.

Doane, George B., III↗
NASA NTRSDate not supplied

Multi-angle Imaging SpectroRadiometer

MISR views the sunlit Earth simultaneously at nine widely spaced angles and provides ongoing global coverage with high spatial detail. Its imagery is carefully calibrated to provide accurate measures of the brightness, contrast, and color of reflected sunlight. MISR provides new types of information for scientists studying Earth's climate, such as the regional and global distribution of different types of atmospheric particles and aerosols. The change in reflection at different view angles provides the means to distinguish aerosol types, cloud forms, and land surface cover. Combined with stereoscopic techniques, this enables construction of 3-D cloud models and estimation of the total amount of sunlight reflected by Earth's diverse environments. MISR was built for NASA by the Jet Propulsion Laboratory (JPL) in Pasadena, California. It is part of NASA's first Earth Observing System (EOS) spacecraft, the Terra spacecraft, which was launched into polar orbit from Vandenberg Air Force Base on December 18, 1999. MISR has been continuously providing data since February 24, 2000. [Mission Objectives] The MISR instrument acquires systematic multi-angle measurements for global monitoring of top-of-atmosphere and surface albedos and for measuring the shortwave radiative properties of aerosols, clouds, and surface scenes in order to characterize their impact on the Earth's climate. The Earth's climate is constantly changing -- as a consequence of both natural processes and human activities. Scientists care a great deal about even small changes in Earth's climate, since they can affect our comfort and well-being, and possibly our survival. A few years of below-average rainfall, an unusually cold winter, or a change in emissions from a coal-burning power plant, can influence the quality of life of people, plants, and animals in the region involved. The goal of NASA's Earth Observing System (EOS) is to increase our understanding of the climate changes that are occurring on our planet, and the reasons for these changes, so we are better equipped to anticipate and prepare for the future. The MISR instrument is a part of EOS. Its role is to measure the amount of sunlight scattered in different directions under natural conditions. These measurements will help quantify the amount of solar energy that heats the Earth's surface and atmosphere, and the changes that occur in them over the lifetime of the MISR instrument. From the MISR observations, we are also learning more about those components of the Earth's environment that scatter sunlight: particles in the atmosphere, the planet's surface, and clouds. MISR monitors changes in surface reflection properties, in atmospheric aerosol content and composition, and in cloudiness. Scientists use these data to study land use changes, air pollution, volcanic eruptions, as well as processes such as desertification, deforestation, and soil erosion. As part of the EOS program, computer models that predict future climate will be improved by the results of these studies. [Temporal_Coverage: Start_Date=2000-02-24; Stop_Date=] [Spatial_Coverage: Southernmost_Latitude=-90; Northernmost_Latitude=90; Westernmost_Longitude=-180; Easternmost_Longitude=180].

Diner, David J.↗
NASA NTRS2008

Boundary Layer Receptivity due to Roughness and Freestream Sound for Supersonic Flows over Axisymmetric Cones

The receptivity of supersonic boundary layers due to the interaction of a two-dimensional acoustic wave with a three-dimensional roughness is numerically investigated over a 5-degree sharp tipped straight cone at a free stream Mach number of 3.5 and a high Reynolds number of 10(exp 6)/inch. Both the steady and unsteady solutions are obtained by solving the full Navier-Stokes equations using the fifth-order accurate weighted essentially non-oscillatory (WENO) scheme for spatial discretization and using third-order total-variation-diminishing (TVD) Runge-Kutta scheme for temporal integration. The simulations are performed with different roughness heights and roughness locations. The steady computations with the roughness reveal that the perturbations induced by the roughness remain very close to the wall and do not penetrate to the outer part of the boundary layer. The effects of the roughness persist for about 10 boundary layer thicknesses downstream of each element. The unsteady simulation shows the generation of unstable three-dimensional first mode disturbances within a short distance from the roughness. However, the amplitudes of the instability waves are very small near the neutral point, on the order of 10(exp -4) compared to the free stream disturbances. The receptivity coefficients due to the roughness elements located very close to the tip are one order of magnitude smaller than those due to the roughness elements located closer to the neutral point.

Balakumar, P.↗
NASA NTRS2019

Constraining Stellar Parameters and Atmospheric Dynamics of the Carbon AGB star V Oph

Molecules and dust produced by the atmospheres of cool evolved stars contribute to a significant amount of the total material found in the interstellar medium. To understand the mechanism behind the mass loss of these stars, it is of pivotal importance to investigate the structure and dynamics of their atmospheres. Our goal is to verify if the extended molecular and dust layers of the carbon-rich asymptotic giant branch (AGB) star V Oph, and their time variations, can be explained by dust-driven winds triggered by stellar pulsation alone, or if other mechanisms are in play. We model V Oph mid-infrared interferometric VLTI-MIDI data (8–13 μm), at phases 0.18, 0.49, and 0.65, together with literature photometric data, using the latest-generation self-consistent dynamic atmosphere models for carbon-rich stars: DARWIN. We determine the fundamental stellar parameters: T(eff) = 2600 K, L(bol) = 3585 L⊙, M = 1.5 M⊙, C/O = 1.35, Mass flow rate= 2.50 x 10(exp -6) M⊙/yr. We calculate the stellar photospheric radii at the three phases: 479, 494, 448 R⊙; and the dust radii: 780, 853, 787 R⊙. The dynamic models can fairly explain the observed N-band visibility and spectra, although there is some discrepancy between the data and the models, which is discussed in the text. We discuss the possible causes of the temporal variations of the outer atmosphere, deriving an estimate of the magnetic field strength, and computing upper limits for the Alfvén waves velocity. In addition, using period–luminosity sequences, and interferometric modeling, we suggest V Oph as a candidate to be reclassified as a semi-regular star.

Gioia Rau↗
DOE OSTI2024

Alternating Direction Decomposition with Strong Bounding and Convexification (ADDSBC) for Solving Security Constrained AC Unit Commitment Problems

This project aims to develop efficient and robust computational methods for solving the security-constrained unit commitment and alternating current optimal power flow problem (SC-UC-ACOPF). The SC-UC-ACOPF problem is at the center of the short-term operation of the U.S. Power Grid. It is solved every week, every day, and every 10 minutes to plan for the optimal action of electricity generation and consumption by minimizing the generation cost and maintaining power system reliability against potential disruptions of equipment failures. In mathematical terms, SC-UC-ACOPF is a challenging large-scale mixed-integer nonlinear optimization model. This means that the decisions involve both discrete variables, e.g. the turning on and off of generators and switching of transmission lines and transformers, and continuous decisions, e.g. the amount of energy generated by each generator and the power flows in the power grid. The physics of the power flow is described by nonlinear equations involving real and reactive power and bus voltages. Another key feature is the large number of contingencies, i.e. the system needs to stay reliable in face of failure of any one equipment, such as transmission lines and generators. The U.S. power grids are extremely complicated and large scale with more than 5,000 generators, 50,000 buses, and 100,000 high-voltage transmission lines, making the SC-UC-ACOPF a very large-scale computation challenge. The research developed in this project aims to solve the SC-UC-ACOPF problems in the three timescales, i.e. weekly, daily, and every 10-min. The proposed computational methods are built on a principled algorithmic approach of decomposition and penalization. More specifically, the algorithm develops spatial and temporal decomposition by exploiting the strong temporal coupling and weak spatial coupling of the UC problem and the complementary feature, i.e. weak temporal coupling and strong spatial coupling of the ACOPF problem. The algorithm also leverages recent progresses in strong convex relaxation of ACOPF. A unique feature of the proposed approach is that it generates a valid, global upper bound on the optimal maximum profit. In this way, a global optimality gap is available to measure the quality of the solution. To further speed up computation, the research team has developed a plethora of effective heuristics to strengthen the iterative penalty-based decomposition framework. For instance, a heuristic is developed to construct inner approximations of the time coupling constraints within the time decoupled problems. Contingencies are pre-screened and low-rank matrix computation is exploited to find the almost unique solution to each contingency. A novel heuristic for line switching is proposed and tested with positive impacts on instances where line switching is beneficial. Taking a systematic approach and carefully handling every detail of the problem pays off. The TIM-GO’s performance throughout the trials and the final event was stellar. TIM-GO garnered the second highest total prize money and is ranked in the top three positions across all categories of comparison.

97 MATHEMATICS AND COMPUTING↗
NASA NTRS2011

Runtime Verification with State Estimation

We introduce the concept of Runtime Verification with State Estimation and show how this concept can be applied to estimate theprobability that a temporal property is satisfied by a run of a program when monitoring overhead is reduced by sampling. In such situations, there may be gaps in the observed program executions, thus making accurate estimation challenging. To deal with the effects of sampling on runtime verification, we view event sequences as observation sequences of a Hidden Markov Model (HMM), use an HMM model of the monitored program to "fill in" sampling-induced gaps in observation sequences, and extend the classic forward algorithm for HMM state estimation (which determines the probability of a state sequence, given an observation sequence) to compute the probability that the property is satisfied by an execution of the program. To validate our approach, we present a case study based on the mission software for a Mars rover. The results of our case study demonstrate high prediction accuracy for the probabilities computed by our algorithm. They also show that our technique is much more accurate than simply evaluating the temporal property on the given observation sequences, ignoring the gaps.

sampling↗
DOE OSTI2024

Forecasting high-dimensional spatio-temporal systems from sparse measurements

This paper introduces a new neural network architecture designed to forecast high-dimensional spatio-temporal data using only sparse measurements. The architecture uses a two-stage end-to-end framework that combines neural ordinary differential equations (NODEs) with vision transformers. Initially, our approach models the underlying dynamics of complex systems within a low-dimensional space; and then it reconstructs the corresponding high-dimensional spatial fields. Many traditional methods involve decoding high-dimensional spatial fields before modeling the dynamics, while some other methods use an encoder to transition from high-dimensional observations to a latent space for dynamic modeling. In contrast, our approach directly uses sparse measurements to model the dynamics, bypassing the need for an encoder. This direct approach simplifies the modeling process, reduces computational complexity, and enhances the efficiency and scalability of the method for large datasets. We demonstrate the effectiveness of our framework through applications to various spatio-temporal systems, including fluid flows and global weather patterns. Although sparse measurements have limitations, our experiments reveal that they are sufficient to forecast system dynamics accurately over long time horizons. Our results also indicate that the performance of our proposed method remains robust across different sensor placement strategies, with further improvements as the number of sensors increases. This robustness underscores the flexibility of our architecture, particularly in real-world scenarios where sensor data is often sparse and unevenly distributed.

97 MATHEMATICS AND COMPUTING↗
DOE OSTI2024

Spatio–Temporal Machine Learning for Regional to Continental Scale Terrestrial Hydrology

Integrated hydrologic models can simulate coupled surface and subsurface processes but are computationally expensive to run at high resolutions over large domains. Here we develop a novel deep learning model to emulate subsurface flows simulated by the integrated ParFlow–CLM model across the contiguous US. We compare convolutional neural networks like ResNet and UNet run autoregressively against our novel architecture called the Forced SpatioTemporal RNN (FSTR). The FSTR model incorporates separate encoding of initial conditions, static parameters, and meteorological forcings, which are fused in a recurrent loop to produce spatiotemporal predictions of groundwater. We evaluate the model architectures on their ability to reproduce 4D pressure heads, water table depths, and surface soil moisture over the contiguous US at 1 km resolution and daily time steps over the course of a full water year. The FSTR model shows superior performance to the baseline models, producing stable simulations that capture both seasonal and event–scale dynamics across a wide array of hydroclimatic regimes. The emulators provide over 1,000× speedup compared to the original physical model, which will enable new capabilities like uncertainty quantification and data assimilation for integrated hydrologic modeling that were not previously possible. Our results demonstrate the promise of using specialized deep learning architectures like FSTR for emulating complex process–based models without sacrificing fidelity.

54 ENVIRONMENTAL SCIENCES↗
NASA NTRSDate not supplied

Rolling Horizon with K-Position Search Method for Strategic Deconfliction of Package Delivery UAS

In this research, the strategic deconfliction of unmanned aircraft systems for an urban package delivery environment with two depots and multiple drop-off locations is studied. This research aims to formulate a mathematical model to compute both the departure sequence and scheduled time of departure for each unmanned aircraft system at a depot, considering temporal constraints at en-route crossing waypoints and depots for strategic deconfliction. However, the problem formulation results in an NP-hard mixed-integer nonlinear programming problem for the global optimal solution, so instead, a "rolling horizon with𝑘-position search"heuristic method is developed. The simulation studies show that an increase in the value of𝑘(the parameter used to determine the size of the local neighborhood) reduces the average ground delay at the cost of an increase in the computation time for a given problem size. The study also shows an order of magnitude increase in the maximum number of flights scheduled with the integration of rolling horizon (time decomposition) compared to those without the integration of rolling horizon in the heuristic algorithm for a given computation time cut off.

UTM↗
DOE OSTI2024

Performance Improvements of the Griffin Solvers in FY24

The Griffin code is a MOOSE-based reactor physics application jointly developed by Idaho National Laboratory and Argonne National Laboratory under the Department of Energy Office of Nuclear Energy Nuclear Energy Advanced Modeling and Simulation Program. This fiscal year, we have made significant efforts to improve the performance of transport solver options and cross-section generation for the efficient use of Griffin in advanced reactor applications. For the HFEM-PN solver, the residual evaluations of HFEM kernels were optimized by utilizing the pre- computed averaged cross sections for individual elements. Numerical integration involving the evaluation of basis functions at quadrature points was bypassed by facilitating precomputed element mass matrices for response matrices. Red-black iterations were improved by introducing a new generalized minimum residual based solver. The memory usage of response matrix storage was significantly reduced by applying basis function rotations on interfaces and calculating volumetric odd-parity moments on the fly. Additionally, the adjoint flux and transient calculation capabilities of the HFEM-PN solver were successfully implemented and verified using the TWIGL benchmark problem. For the DFEM-SN solver, memory footprint and computation time were significantly reduced by not treating angular flux vectors as the MOOSE nonlinear system vectors. Specifically for IQS, scalar adjoint weighting was introduced to further eliminate angular adjoint flux storage in the MOOSE auxiliary system. It was demonstrated through the three-dimensional Advanced Burner Test Reactor core problem that the memory usage for transient calculations with the IQS method was reduced by over 7.5× compared to before the optimizations. For the self-shielding application programming interface, a new double-heterogeneity treatment method, named the Bell Function-Based Analytic Two-Region Slowing Down Method, was developed to efficiently flux-volume homogenize TRISO particles with the matrix. Additionally, optimizations were made to hyper- fine group (HFG) slowing down calculations by pretabulating collision probability coefficients and grouping isotopes, significantly reducing the computational time for calculating scattering sources per HFG. Lastly, the pin power reconstruction module was extended to account for temporal behavior in a microreactor analysis problem, specifically for a control drum transient. Verification tests for each of these improvements demonstrated significant performance enhancements and memory reduction.

22 - GENERAL STUDIES OF NUCLEAR REACTORS↗
DOE OSTI2025

Compactly‐Supported Nonstationary Kernels for Computing Exact Gaussian Processes on Big Data

The Gaussian process (GP) is a widely used method for analyzing large-scale data sets, including spatio-temporal measurements of nonlinear processes that are now commonplace in the environmental sciences. Traditional implementations of GPs involve stationary kernels (also termed covariance functions) that limit their flexibility, and exact methods for inference that prevent application to data sets with more than about 10,000 points. Modern approaches to address stationarity assumptions generally fail to accommodate large data sets, while all attempts to address scalability focus on approximating the Gaussian likelihood, which can involve subjectivity and lead to inaccuracies. In this work, we explicitly derive an alternative kernel that can discover and encode both sparsity and nonstationarity. We embed the kernel within a fully Bayesian GP model and leverage high-performance computing resources to enable the analysis of massive data sets. We demonstrate the favorable performance of our novel kernel relative to existing exact and approximate GP methods across a variety of synthetic data examples. Furthermore, we conduct space–time prediction based on more than 1 million measurements of daily maximum temperature and verify that our results outperform state-of-the-art methods in the Earth sciences. More broadly, having access to exact GPs that use ultra-scalable, sparsity-discovering, nonstationary kernels allows GP methods to truly compete with a wide variety of machine learning methods.

Gaussian processes↗
NASA NTRS2001

Grain Properties of Comet C/1995 O1 (Hale-Bopp) Deduced Through Computational Techniques

We present the computational analysis of the 7.6 - 13.2 micrometer infrared (IR) spectrophotometry (R approximately equal to 120) of comet C/1995 O1 (Hale-Bopp) in conjunction with concurrent observations which extend the spectral energy distribution from the near-infrared to far-infrared wavelengths. The observations include temporal epochs pre-perihelion, (1996 October UT and 1997 February UT), near perihelion (1997 April UT), and postperihelion (1997 June UT). Through the computational modeling of small amorphous carbon, and crystalline and amorphous silicate grains in Hale-Bopp's coma, we find that as the comet approached perihelion, the grain size distribution (the Hanner modified power law) steepened (N = 3.4 pre-perihelion, to N = 3.7 near and post-perihelion) along with an increase in the fractal porosity of larger (greater than 1 micrometer) grains. The peak of the grain size distribution remained constant (ap = 0.2 micrometer) at each epoch. We attribute the emergence of the 9.3 micrometer peak near perihelion to crystalline orthopyroxeno grains released from inside the nucleus. Crystalline silicates (olivine and orthopyroxene) make up about 30% (by mass) of the submicron sized (less than 1 micrometer) dust grains in Hale-Bopp's coma during each epoch.

Harker, David E.↗
DOE OSTI2026

ObstacleSense: Low-Power Neuromorphic Vision for Corridor Obstacle Awareness in Low-Level ADAS

The automotive industry’s pursuit of Level 5 autonomy is constrained by substantial perception-compute power requirements, often reaching 1, 000 + watts in full autonomy stacks. Reducing this energy burden requires rethinking perception not only at the high-end autonomy level, but also at the foundational Advanced Driver Assistance Systems (ADAS) level where low-power, safety-critical sensing can have broad impact. Neuromorphic vision provides a promising starting point: HD Dynamic Vision Sensors (DVS) can operate below 100 mW at the sensor level by reporting only asynchronous brightness changes. However, low-power sensing alone is insufficient if downstream perception reintroduces dense, energy-intensive computation. In particular, many event-driven object-detection pipelines still rely on CNN backbones, while purely spiking alternatives often trade away accuracy or ignore deployment constraints. We introduce ObstacleSense, a highly compact, CNN-free hybrid ANN–SNN framework for Level 0–1 forward-corridor obstacle awareness. Instead of performing full-scene object detection with a convolutional feature backbone, ObstacleSense targets the safety-critical question of whether the ego corridor is occupied and how far the nearest obstacle is. The architecture combines polarity-conditioned event encoding, lightweight temporal spiking dynamics, axial spatial mixing, and coarse-to-fine range estimation within a regular fixed-grid compute pattern. This design avoids the dense CNN backbone commonly used in event-based detection while maintaining a small state footprint suitable for eventual small-FPGA deployment. Before hardware mapping, we evaluate the software implementation using a model-side power proxy derived from MACs, weight and activation traffic, and spiking state updates under shared FP16 assumptions. On simulated CARLA event corpora, the deployment-oriented model achieves 0.9464 objectness F1, 0.9978 grid-level mAP, and 0.8987 m distance Mean Absolute Error at an estimated 1.92 mW proxy cost, while maintaining performance on unseen generalization test sequences.

Johnson-Scott, Zac [ORNL]↗