Engineering topics
Mandrake, Lukas
Publications and source records attributed to Mandrake, Lukas.
Levels of On-Board Autonomy: Application of ML to Mission Architecture
No abstract available
A New BRDF Model to Reduce Biases in Orbiting Carbon Observatory-2 (OCO-2) Retrievals
No abstract available
BRDF models to reduce biases in Orbiting Carbon Observatory-2 (OCO-2) retrievals
UNKNOWN
Scatter-Reducing Sounding Filtration Using a Genetic Algorithm and Mean Monthly Standard Deviation
Retrieval algorithms like that used by the Orbiting Carbon Observatory (OCO)-2 mission generate massive quantities of data of varying quality and reliability. A computationally efficient, simple method of labeling problematic datapoints or predicting soundings that will fail is required for basic operation, given that only 6% of the retrieved data may be operationally processed. This method automatically obtains a filter designed to reduce scatter based on a small number of input features. Most machine-learning filter construction algorithms attempt to predict error in the CO2 value. By using a surrogate goal of Mean Monthly STDEV, the goal is to reduce the retrieved CO2 scatter rather than solving the harder problem of reducing CO2 error. This lends itself to improved interpretability and performance. This software reduces the scatter of retrieved CO2 values globally based on a minimum number of input features. It can be used as a prefilter to reduce the number of soundings requested, or as a post-filter to label data quality. The use of the MMS (Mean Monthly Standard deviation) provides a much cleaner, clearer filter than the standard ABS(CO2-truth) metrics previously employed by competitor methods. The software's main strength lies in a clearer (i.e., fewer features required) filter that more efficiently reduces scatter in retrieved CO2 rather than focusing on the more complex (and easily removed) bias issues.
Informing Climate Retrieval Development Using Data Mining or How to Help Scientist Draw Better Lines
No abstract available
Automatic Data Filter Customization Using a Genetic Algorithm
This work predicts whether a retrieval algorithm will usefully determine CO2 concentration from an input spectrum of GOSAT (Greenhouse Gases Observing Satellite). This was done to eliminate needless runtime on atmospheric soundings that would never yield useful results. A space of 50 dimensions was examined for predictive power on the final CO2 results. Retrieval algorithms are frequently expensive to run, and wasted effort defeats requirements and expends needless resources. This algorithm could be used to help predict and filter unneeded runs in any computationally expensive regime. Traditional methods such as the Fischer discriminant analysis and decision trees can attempt to predict whether a sounding will be properly processed. However, this work sought to detect a subsection of the dimensional space that can be simply filtered out to eliminate unwanted runs. LDAs (linear discriminant analyses) and other systems examine the entire data and judge a "best fit," giving equal weight to complex and problematic regions as well as simple, clear-cut regions. In this implementation, a genetic space of "left" and "right" thresholds outside of which all data are rejected was defined. These left/right pairs are created for each of the 50 input dimensions. A genetic algorithm then runs through countless potential filter settings using a JPL computer cluster, optimizing the tossed-out data s yield (proper vs. improper run removal) and number of points tossed. This solution is robust to an arbitrary decision boundary within the data and avoids the global optimization problem of whole-dataset fitting using LDA or decision trees. It filters out runs that would not have produced useful CO2 values to save needless computation. This would be an algorithmic preprocessing improvement to any computationally expensive system.
Mixture-Tuned, Clutter Matched Filter for Remote Detection of Subpixel Spectral Signals
Mapping localized spectral features in large images demands sensitive and robust detection algorithms. Two aspects of large images that can harm matched-filter detection performance are addressed simultaneously. First, multimodal backgrounds may thwart the typical Gaussian model. Second, outlier features can trigger false detections from large projections onto the target vector. Two state-of-the-art approaches are combined that independently address outlier false positives and multimodal backgrounds. The background clustering models multimodal backgrounds, and the mixture tuned matched filter (MT-MF) addresses outliers. Combining the two methods captures significant additional performance benefits. The resulting mixture tuned clutter matched filter (MT-CMF) shows effective performance on simulated and airborne datasets. The classical MNF transform was applied, followed by k-means clustering. Then, each cluster s mean, covariance, and the corresponding eigenvalues were estimated. This yields a cluster-specific matched filter estimate as well as a cluster- specific feasibility score to flag outlier false positives. The technology described is a proof of concept that may be employed in future target detection and mapping applications for remote imaging spectrometers. It is of most direct relevance to JPL proposals for airborne and orbital hyperspectral instruments. Applications include subpixel target detection in hyperspectral scenes for military surveillance. Earth science applications include mineralogical mapping, species discrimination for ecosystem health monitoring, and land use classification.
Histogrammatic Method for Determining Relative Abundance of Input Gas Pulse
To satisfy the Major Constituents Analysis (MCA) requirements for the Vehicle Cabin Atmosphere Monitor (VCAM), this software analyzes the relative abundance ratios for N2, O2, Ar, and CO2 as a function of time and constructs their best-estimate mean. A histogram is first built of all abundance ratios for each of the species vs time. The abundance peaks corresponding to the intended measurement and any obfuscating background are then separated via standard peak-finding techniques in histogram space. A voting scheme is then used to include/exclude this particular time sample in the final average based on its membership to the intended measurement or the background population. This results in a robust and reasonable estimate of the abundance of trace components such as CO2 and Ar even in the presence of obfuscating backgrounds internal to the VCAM device. VCAM can provide a means for monitoring the air within the enclosed environments, such as the ISS (International Space Station), Crew Exploration Vehicle (CEV), a Lunar Habitat, or another vehicle traveling to Mars. Its miniature pre-concentrator, gas chromatograph (GC), and mass spectrometer can provide unbiased detection of a large number of organic species as well as MCA analysis. VCAM s software can identify the concentration of trace chemicals and whether the chemicals are on a targeted list of hazardous compounds. This innovation s performance and reliability on orbit, along with the ground team s assessment of its raw data and analysis results, will validate its technology for future use and development.
Major Constituents Analysis for the Vehicle Cabin Atmosphere Monitor
Vehicle Cabin Atmosphere Monitor (VCAM) can provide a means for monitoring the air within enclosed environments such as the International Space Station, the Crew Exploration Vehicle (CEV), a Lunar habitat, or another vehicle traveling to Mars. The software processes a sum total spectra (counts vs. mass channel) with the intention of computing abundance ratios for N2, O2, CO2, Ar2, and H2O. A brute-force powerset expansion compares a library of expected mass lines with those found within the data. Least squares error is combined with a penalty term for using small peaks.
Adaption of the AMDIS Method to Flight Status on the VCAM Instrument
Software has been developed to function onboard the International Space Station (ISS) to help safeguard human health by detecting compounds of concern in the cabin atmosphere, both in identity and concentration.
Quantification of Trace Chemicals Using Vehicle Cabin Atmosphere Monitor
A system to monitor the concentrations of trace chemicals in cabin atmosphere is one of the most critical components in long-duration human flight missions. The Vehicle Cabin Atmosphere Monitor (VCAM) is a miniature gas chromatograph mass spectrometer system to be used to detect and quantify trace chemicals in the International Space Station. We developed an autonomous computational process to quantify trace chemicals for use in VCAM. The process involves the design of a measured signal quantification scheme, the construction of concentration curves (i.e. the relationship between concentration and ion count measured by VCAM), the decision rule of applying high- or low-gain concentration curves, and the detection of saturation, low-signals, and outliers. When the developed quantification process is applied, the average errors of concentration for most of trace chemicals are found to be between 14% and 66%.