The Machine Learning for Correlated Intelligence Laboratory Directed Research & Development (LDRD) Project explored competing a variety of machine learning (ML) classification techniques against a known, open source dataset through the use of a rapid and automated algorithm research & development (RD) infrastructure. This approach relied heavily on creating an infrastructure in which to provide a pipeline for automatic target recognition (ATR) ML algorithm competition. Results are presented for nine ML classifiers against a primary dataset using the pipeline infrastructure developed for this project. New approaches to feature set extraction are presented and discussed as well.
Low signal-to-noise data processing algorithms for improved detection, tracking, discrimination and situational threat assessment are a key research challenge. As sensor technologies progress, the number of pixels will increase significantly. This will result in increased resolution, which could improve object discrimination, but unfortunately, will also result in a significant increase in the number of potential targets to track. Many tracking techniques, like multi-hypothesis trackers, suffer from a combinatorial explosion as the number of potential targets increase. As the resolution increases, the phenomenology applied towards detection algorithms also changes. For low resolution sensors, blob tracking is the norm. For higher resolution data, additional information may be employed in the detection and classification steps. The most challenging scenarios are those where the targets cannot be fully resolved, yet must be tracked and distinguished for neighboring closely spaced objects. Tracking vehicles in an urban environment is an example of such a challenging scenario. This report evaluates several potential tracking algorithms for large-scale tracking in an urban environment. The algorithms considered are: random sample consensus (RANSAC), Markov chain Monte Carlo data association (MCMCDA), tracklet inference from factor graphs, and a proximity tracker. Each algorithm was tested on a combination of real and simulated data and evaluated against a common set of metrics.
Low signal-to-noise data processing algorithms for improved detection, tracking, discrimination and situational threat assessment are a key research challenge. As sensor technologies progress, the number of pixels will increase signi cantly. This will result in increased resolution, which could improve object discrimination, but unfortunately, will also result in a significant increase in the number of potential targets to track. Many tracking techniques, like multi-hypothesis trackers, suffer from a combinatorial explosion as the number of potential targets increase. As the resolution increases, the phenomenology applied towards detection algorithms also changes. For low resolution sensors, "blob" tracking is the norm. For higher resolution data, additional information may be employed in the detection and classfication steps. The most challenging scenarios are those where the targets cannot be fully resolved, yet must be tracked and distinguished for neighboring closely spaced objects. Tracking vehicles in an urban environment is an example of such a challenging scenario. This report evaluates several potential tracking algorithms for large-scale tracking in an urban environment.
We derive from first principles a mathematical physics model useful for understanding nonlinear optical propagation (including filamentation). All assumptions necessary for the development are clearly explained. We include the Kerr effect, Raman scattering, and ionization (as well as linear and nonlinear shock, diffraction and dispersion). We explain the phenomenological sub-models and each assumption required to arrive at a complete and consistent theoretical description. The development includes the relationship between shock and ionization and demonstrates why inclusion of Drude model impedance effects alters the nature of the shock operator. Unclassified Unlimited Release
Future pulsed power systems may rely on linear transformer driver (LTD) technology. The LTD's will be the building blocks for a driver that can deliver higher current than the Z-Machine. The LTD's would require tens of thousands of low inductance ( %3C 85nH), high voltage (200 kV DC) switches with high reliability and long lifetime ( 10 4 shots). Sandia's Z-Machine employs 36 megavolt class switches that are laser triggered by a single channel discharge. This is feasible for tens of switches but the high inductance and short switch life- time associated with the single channel discharge are undesirable for future machines. Thus the fundamental problem is how to lower inductance and losses while increasing switch life- time and reliability. These goals can be achieved by increasing the number of current-carrying channels. The rail gap switch is ideal for this purpose. Although those switches have been extensively studied during the past decades, each effort has only characterized a particular switch. There is no comprehensive understanding of the underlying physics that would allow predictive capability for arbitrary switch geometry. We have studied rail gap switches via an extensive suite of advanced diagnostics in synergy with theoretical physics and advanced modeling capability. Design and topology of multichannel switches as they relate to discharge dynamics are investigated. This involves electrically and optically triggered rail gaps, as well as discrete multi-site switch concepts.
To extend the backlighting capabilities for Sandia's Z-Accelerator, Z-Petawatt, a laser which can provide laser pulses of 500 fs length and up to 120 J (100TW target area) or up to 450 J (Z / Petawatt target area) has been built over the last years. The main mission of this facility focuses on the generation of high energy X-rays, such as tin Ka at 25 keV in ultra-short bursts. Achieving 25 keV radiographs with decent resolution and contrast required addressing multiple problems such as blocking of hot electrons, minimization of the source, development of suitable filters, and optimization of laser intensity. Due to the violent environment inside of Z, an additional very challenging task is finding massive debris and radiation protection measures without losing the functionality of the backlighting system. We will present the first experiments on 25 keV backlighting including an analysis of image quality and X-ray efficiency.
To extend the backlighting capabilities for Sandia's Z-Accelerator, Z-Petawatt, a laser which can provide laser pulses of 500 fs length and up to 120 J (100TW target area) or up to 450 J (Z/Petawatt target area) has been built over the last years. The main mission of this facility focuses on the generation of high energy X-rays, such as tin K{alpha} at 25 keV in ultra-short bursts. Achieving 25 keV radiographs with decent resolution and contrast required addressing multiple problems such as blocking of hot electrons, minimization of the source, development of suitable filters, and optimization of laser intensity. Due to the violent environment inside of Z, an additional very challenging task is finding massive debris and radiation protection measures without losing the functionality of the backlighting system. We will present the first experiments on 25 keV backlighting including an analysis of image quality and X-ray efficiency.
The Space Shuttle Program requires on-orbit inspection of the thermal protection system which covers the Orbiter spacecraft, including the critical leading-edge surfaces. A scannerless ladar system mounted on a 50-foot boom extension of the robotic arm provides this capability. This paper describes the sensor and ground processing system, which were developed by Sandia National Laboratories to meet the requirements of the Return to Flight mission in July of 2005. Mission operations for this sensor system are also reviewed.