SEARCH · Engineering Papers
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.
Automated planning and scheduling, including automated path planning, has been integrated with an internet-based distributed operations system for planetary rover operations.
It is well known that the Earth has an ongoing problem with orbiting space debris. Some Earth orbiting missions have regular warnings of close approaches with debris or other satellites. At Mars and the Moon, due to the growing number of orbiter missions and the current inability to track orbital debris in these environments, the creation of a hazardous debris field must be avoided because a debris field would greatly complicate both existing and future operations. Work at the Jet Propulsion Laboratory in the area of automated spacecraft conjunction assessment at Mars and the Moon has been conducted over the past six years using a process called "MADCAP" ("Multimission Automated Deepspace Conjunction Assessment Process"). A paper introducing this work was presented at Space Ops in Stockholm, Sweden in 2012. In that inaugural paper, the then current state of operations was presented along with a number of items that were identified for potential future work. The fundamental design concepts of MADCAP have not materially changed in the last five years, however, since 2012 a number of the changes to MADCAP identified in the previous paper have been implemented. Some other previously planned work has not progressed appreciably; several of these items remain on a "parking lot" list. In addition to the items that were listed as prospective future work, JPL's Mars/Moon conjunction assessment efforts have also been extended in a few unplanned but important areas. This follow-up paper will provide a five year update on MADCAP operations at Mars and the Moon.
The Mars Science Laboratory (MSL) Telecom Operations Team at the Jet Propulsion Laboratory (JPL) has implemented a machine learning system in order to automate the anomaly detection process as a part of daily operations. Machine learning enables reliable detection of anomalies in Telecom-related telemetry and automated reporting of Telecom subsystem status, resulting in an 90% reduction in team workload and improved anomaly detection reliability. At present, machine learning methods are used to detect: 1. Anomalous long-term trends in telemetry data 2. Anomalous time-domain evolution of telemetry values Both types of anomalies pose their own unique challenges that are addressed in different ways. In the first case, long term trending of daily minima, maximum, and mean telemetry values in temperatures, currents, voltages, and radio frequency (RF) power levels is used in addition to hard threshold safety checks to look for changes in long-term equipment health and performance. Long-term trending methods allow for ordinary seasonal variations in these quantities caused by temperature changes over the course of the Martian year while allowing operators to determine whether current performance remains in line with historical values from previous years. Changes in long-term trends can provide important insights into the health and status of the rover's on-board systems as well as valuable early warning if subtle degradation begins to take hold. But while trending of daily statistics is valuable, it does not detect anomalies in the short-term time evolution of data over the course of minutes or hours during a day, and this task is handled with short-term shape analysis. Principal components analysis (PCA) has been found to provide robust detection of short-term anomalies, and several examples of the use of PCA to detect actual anomalous events will be provided here. In using PCA, we use both the percentage of explained variance and also a log likelihood test on the PCA expansion coefficients to flag telemetry data for human review. Previous work in the field of spacecraft anomaly detection includes [1] for MSL and [2] for some other JPL missions.
Design documents are intended to outline the goalsof a system or project, which are utilized in the creation ofspecific software requirements. At the NASA Jet PropulsionLaboratory, California Institute of Technology, Functional DesignDescription (FDD) documents describe the scope of theproject and reflect the design and implementation of the system.The specifications in the document are not explicitly writtenas requirements, though these guidelines must be reflected inthe official software requirements. In this work we present afully automatic approach to extracting software requirementsfrom design documents as well as comparing the extractedrequirements to those that exist in the official software requirementdatabase. We do this through (1) sentence extractionfrom the design document, (2) the incorporation of coreferenttext, and (3) aligning the extracted text to the official softwarerequirements. Via natural language processing and informationretrieval techniques, our system results in an automated processthat ensures that the specifications in the design document resultin official software requirements. We find that extraction ofimperatives results in a recall rate of 0.73 and the TF-IDF cosinesimilarity metric is shown to be a useful and successful way tocompare requirements.Though there has been recent work investigating the usefulnessof natural language processing techniques in requirement engineering,this has not been made use of in the aerospace industry.Aerospace requirement engineering is a field particularly ripefor this type of innovation because these techniques can bothautomate some of needlessly manual work and contribute toaerospace safety practices by identifying issues that a humanmay miss. We present the first fully automated approach thatextracts requirements from a design document and comparesthem to a database, and use these findings as encouragementfor future work that makes use of natural language processingtechniques in aerospace requirement engineering.
As the Mars Curiosity Rover transmits data to the JPL Ground Data System (GDS), it frequently observes data loss and corruption, requiring re-transmits from the rover and Ground Data System Analysts (GDSA) to monitor the downlink process. As new missions are launched, the GDSA team redistributes analysts to these new missions, causing shortages in previous missions. The GDSA team can significantly benefit from the automation and optimization of the downlink process of telemetry data. In fact, there is a need for a better understanding of why the data is corrupted, so that the GDSA team can best determine the root cause of the issues in the GDS. This paper presents machine learning and deep learning based approaches to automate and optimize the detection of data loss. We first created a pipeline to automatically accumulate data from the telemetry databases (MAROS, Telemetry Data Storage, and GDS Elastic Search Database) in the downlink process. With our newly created datasets, we perform feature selection to supplement the GDSA understanding of the downlink process and provide supplemental analysis on the importance of different features. We implement various machine learning and deep learning based models, including support vector machines, ensemble methods, and deep neural networks and evaluate their accuracies in identifying whether a downlink process is complete or incomplete. We utilize fast hyperparameter optimization methods that allow our models to quickly be re-trained, allowing them to quickly be tuned and optimized on daily incoming data in real time. This hyperparameter optimization also allows our methods to be quickly integrated into other JPL missions. Our results show that our best-performing machine learning and deep learning based models outperform the existing GDSA detection software by 6 accuracy points and can aid analysts by providing insights into the data accountability problem. Since these various machine learning and deep learning approaches vary significantly in interpretability, we provide a discussion on the tradeoffs between their performance and trustworthiness in helping detect issues in data transmission.
Qualification and certification of safety critical parts is a hurdle to the adoption of metallic additively manufactured components for aerospace vehicle applications. Challenges include variability in part properties due to inconsistent defect distribution and microstructure. Understanding of the process through finite element modeling (FEM), and process control through in-situ monitoring, may result in significant improvements; however, solutions useful to manufacturers will require large volumes of data and automated data utilization. Toward this end, a generalizable automated FEM calibration paradigm is developed. This paradigm leverages existing and novel tools from machine learning and uncertainty quantification to enable the automatic calibration of FEMs without requiring prior knowledge of the model performance across input parameter space, including meshing and solver settings, which can require time consuming manual model probing or cause noisy and inconsistent predictions. The result is a probabilistic distribution of calibrated and validated FEM input parameters targeting measured data.
The Earth Surface Mineral Dust Source InvesTigation (EMIT) is an Earth Ventures-Instrument (EVI-4) mission to map the surface mineralogy of arid dust source regions. EMIT used automated scheduling technology to analyze aspects of the mission design. The automated scheduling technology was used to construct schedules which were then automatically analyzed with respect to science acquired. These analyses can be performed for a range of spacecraft hardware configurations, observation strategies, and science requirements. By studying the effects of changes on the above inputs, better hardware configurations, observation strategies, and science requirements can be formulated. The use of a pointing mirror on EMIT was under consideration, and this analysis aided in determining whether or not to keep it as part of the design of the instrument. Clouds will also have a large impact on the coverage of science targets achievable by the mission. Analysis was done on how clouds could impact the coverage achievable as well as the data volume. This analysis with clouds also aided in determining the coverage criteria for the mission. It was necessary to find a criteria that was achievable with some margin as well as satisfies the science goals of the mission.
The Deep Space Network (DSN) comprises three sites, located in California, Spain, and Australia; each site operates one 70m and multiple 34m antennas that provide communications and navigation services to NASA and international partners. As part of a multiyear upgrade in automation of the network, the DSN has undertaken a series of operations efficiency improvements which has fundamentally changed the operations paradigm of the network. At the start of this effort, each site operated only their local antennas and equipment 24 hours/day, 7 days/week, and each spacecraft activity (link) had a single dedicated Link Control Operator. In its final stage, realized in 2020, each one of the three sites operates the entire network during their local day shift, handing off control to the next site as their day ends, and tracks can be run in a fully automated manner with no operator intervention. This paradigm change has realized significant operations cost savings, but has posed some major challenges along the way. This paper describes the system changes that enabled the operations efficiency improvements, some the challenges, and plans for the future.
For many decades, researchers at NASA Ames Research Center have worked to make the air-transportation system more efficient, predictable, and effective. Since about 2005 one important aspect of this research has been the development of an autonomous system for air-traffic control. This system, known as the Autoresolver, is designed to perform most of the roles that air-traffic controllers perform including ensuring separation between aircraft, creating routes around weather and other avoidance volumes, and sequencing and scheduling aircraft across points in space. The recent, rapid expansion of new aircraft operations and types, including urban air mobility aircraft and small unmanned aerial systems, have only increased the need for highly automated systems to control the predicted traffic demand. This talk will focus on the development of the Autoresolver - from concept to testing. It will also discuss the National Airspace (NAS) Digital Twin simulation platform, created to facilitate rapid testing and improvement of the algorithm and with the hope of proving the automation in a high-fidelity environment. An open question that will be discussed is how to ensure that the system-level emergent behavior of independently developed autonomous algorithms is what is desired.
While many studies have been performed examining Urban Air Mobility (UAM) operations from UAM Maturity Level (UML) UML-1 to UML-4, [1, 2] some uncertainty exists regarding the integration and role of onboard autonomous systems, airspace management systems, ground control and fleet management systems, and how they integrate with vertiport automation systems to ensure safe high-density future operations. One thrust of the Advanced Air Mobility (AAM) High Density Vertiplex (HDV) sub-project is to perform rapid prototyping and assessment of an Urban Air Mobility (UAM) Ecosystem within the terminal operational area to help inform future research investments and technology development. Another thrust within HDV is to perform integration, testing, and safety risk assessments required to acquire operational credit for several NASA small Unmanned Aerial Systems (sUAS) beyond visual line of sight (BVLOS) enabling technologies to expand test capabilities and to expedite technology transfer and ultimate effective usage. Both thrusts leverage sUAS to serve as surrogates for the highly-technologically-similar envisioned UAM aircraft as well as to provide significant contributions to sUAS Part-135 operators. This report provides an overview of the activities accomplished within the Advanced Onboard Automation (AOA) schedule work package of HDV.
Humans are frequently left to “backstop” automated systems, and Human Factors specialists have argued against this for decades with, at best, partial success. What if we took a different tack... and designed to support it? The participants were involved in a recent effort to review and document cases across multiple domains where operators acted as a “failsafe” for automation, intervening in unanticipated situations to maximize success and minimize damage. We defined a “Human As Failsafe” (HAF) incident and then investigated conditions and practices making HAF success more or less likely. Analyzing these historical incidents, we suggested remediation approaches. The project also examined the legal concept of culpability (i.e., when intervention should have happened but didn’t) and proposed a state-machine-based analytic simulation to identify when HAF interventions are plausible. The panel objective will be to briefly present these concepts, but more generally to discuss designing for inevitable HAF events. This presentation with review the concept of a HAT Assistant to support multi-vehicle control of drones in a wildland firefighting context.
NASA's Earth Observing System Data and Information System (EOSDIS) began dataset Digital Object Identifier (DOI) registration in 2012. The number of dataset DOIs registered as of January of 2023 exceeds 11,000. As the research community becomes aware of the importance of sharing data through Open Science and optimizing data reuse through Findability, Accessibility, Interoperability, and Reuse (FAIR) data management principles, datasets are increasingly being cited in scientific publications. When datasets are cited explicitly by DOI within published works, automated methods can be developed for collecting these published works from a variety of bibliometric sources. The coverage of the sources varies, so each source can collect citations that are only available within it. Using major citation databases such as Scopus and Web of Science, the Google Scholar search engine, the CrossRef Open Citation Index, and the dataset DOI registry DataCite, we present an automated workflow for dataset citation collection. By harvesting citations automatically, a citation library is created explicitly linking EOSDIS datasets to publications that cite them. Using Zotero, a free and open-source citation manager, we demonstrate how to access and browse this library by the tags indicating bibliometric sources, dataset DOI, and the dataset archive center. We also demonstrate temporary trends in the number of publications harvested from bibliometric sources.
The Mars 2020 Rover mission is using automated scheduling in two capacities. As of March 2021 a ground-based scheduler automatically schedules rover wake/sleep and preheats to support Mars 2020 daily operations. Included in the ground system is an explainable scheduling tool that allows users in-sight into the schedule developed by the automated scheduler. An onboard scheduler that will allow the rover autonomously better account for variations in execution is in development for use later in the mission.
Recent successes in Integrated Computational Materials Engineering (ICME) have demonstrated the potential in designing fit-for-purpose materials for a given application in a cost and time efficient manner. However, the material design process must contain a level of automation in the material decision process, implementing some optimization algorithms, to truly enable the full benefits of ICME, particularly when considering materials at multiple length/time scales. In this work, we will demonstrate how the GRC ICME schema and Python framework automates a workflow that captures, analyzes, maintains, and disseminates the digital footprint in the context of tailoring resin material at the nanoscale of a woven composite Y-joint at the macroscale for an Aurora D8 double bubble fuselage. This digital footprint incorporates the interaction of both structural digital twins and material twins at various length scales.
The Airborne Collision Avoidance System X (ACAS X) is a next-generation collision avoidance tool developed to support different aircraft types and operations. The rotorcraft variant – referred to as ACAS XR – is designed to accommodate existing helicopter platforms as well as in-development, electric vertical takeoff and landing concepts, which are critical to the emerging concept of operations referred to as Advanced Air Mobility. The fundamental role of ACAS XR is to provide Detect and Avoid (DAA) and/or Collision Avoidance (CA) protection against airborne traffic. DAA alerting and guidance in the context of ACAS XR is caution-level and “suggestive,” and is to be used by the pilot if, and when, they decide to maneuver against an identified threat to DAA “well clear.” The CA alerting, by contrast, is warning-level and “directive,” with the pilot required to comply with the associated guidance to prevent a predicted Near Midair Collision (NMAC). The CA alerts generated by ACAS XR are referred to as Resolution Advisories (RAs) consistent with previous CA systems. Unlike earlier CA systems, ACAS XR issues RAs in the horizontal and vertical dimensions as well as multi-axis RAs (referred to as “Blended” RAs). According to the Minimal Operational Performance Standards of DAA systems for Unmanned Aircraft Systems, maneuvers to comply with RAs may be automated, whereas maneuvers based on DAA alerting assume a manual response. The current presentation discusses two human-in-the-loop simulations that presented rotorcraft pilots with ACAS XR alerts and guidance in eVTOL simulators with varying levels of automation. It also presents overviews of these studies as well as how they assisted live flight tests, which will occur throughout 2023.
Earth’s magnetosphere is continuously impacted by solar wind and interplanetary magnetic field (IMF) disturbances, such as shocks, discontinuities, magnetic clouds and more. Understanding how such disturbances propagate from the Sun and what is their impact on the different magnetospheric domains is key to understanding and forecasting energy transfer from the solar wind to Earth. The large number of overlapping solar wind and magnetospheric missions carrying magnetometers and the recent advances in communications and data storage technologies have enabled an unprecedented quantity of high-fidelity magnetic field data captured by in-situ spacecraft to be available at the click of a button. However, this massive quantity of available data can prove unwieldy for researchers, limiting the identification of interesting phenomena and disturbances to a relatively small percentage of the total dataset. Several techniques have been previously developed for automated identification of specific types of magnetic anomalies, but these methods are typically mission-specific and can be difficult to generalize. We present initial results for a generic method of automated anomaly detection in magnetic field measurements based on dimensionality reduction and unsupervised clustering via machine learning. The benefit of our technique is its high degree of generalizability and flexibility which make it a most useful data survey tool for a wide range of magnetic field datasets. This method can also be applied simultaneously to other observed time-series properties like plasma density, pressure, and velocity for more accurate event identification. Additionally, the application of this method to data captured by multiple spacecraft enables the simultaneous identification of disturbances and the determination of their propagation characteristics. Initial evaluation of this technique has been performed using data from Magnetospheric MultiScale (MMS) and THEMIS-ARTEMIS missions, providing a testbed scenario for the future Heliophysics Environmental and Radiation Measurement Experiment Suite (HERMES) platform instruments that will measure solar wind and IMF properties from lunar orbit onboard the Gateway station.
Arc jet Computer Vision (arcjetCV)[1][2] is a software application built to automate time-resolved recession tracking of edges in test videos, specifically for tracking material recession and the shock-material standoff. This provides a new capability to resolve and validate new physics associated with non-linear processes and an essential step to reduce testing, modeling, and validation uncertainties for heatshield material performance. ArcticCV uses several types of machine learning (convolutional neural net: CNN, decision tree: DT, k-means unsupervised clustering: KM) to automate the video processing pipeline. These include inferring the start/stop of time segments of interest (1D CNN), measuring the time-dependent 2D recession of the material samples (2D CNN, DT), measuring the time-dependent shock standoff distance (2D CNN, DT), and post-processing cleaning of the recession data (KM). The software also provides a graphical user interface for ease of use. The results of using this tool on arc jet videos show non-linear time-dependent effects can be important for certain materials and characterizing certain failure modes.
Arc jet Computer Vision (arcjetCV) is a software application built to automate time-resolved recession tracking of edges in test videos, specifically for tracking material recession and the shock-material standoff. This provides a new capability to resolve and validate new physics associated with non-linear processes and an essential step to reduce testing, modeling, and validation uncertainties for heatshield material performance. ArcticCV uses several types of machine learning (convolutional neural net: CNN, decision tree: DT, k-means unsupervised clustering: KM) to automate the video processing pipeline. These include inferring the start/stop of time segments of interest (1D CNN), measuring the time-dependent 2D recession of the material samples (2D CNN, DT), measuring the time-dependent shock standoff distance (2D CNN, DT), and post-processing cleaning of the recession data (KM). The software also provides a graphical user interface for ease of use. The results of using this tool on arc jet videos show non-linear time-dependent effects can be important for certain materials and characterizing certain failure modes.