Organizational Unit:

Space Systems Design Laboratory (SSDL)

Permanent Link

https://hdl.handle.net/1853/70747

Parent Organization

Organizational Unit

Daniel Guggenheim School of Aerospace Engineering

ArchiveSpace Name Record

https://finding-aids.library.gatech.edu/agents/corporate_entities/1138

Full item page

Publication Search Results

Now showing 1 - 9 of 9

Generalized Minimum-Time Follow-up Approaches Applied to Tasking Electro-Optical Sensor Tasking

(Georgia Institute of Technology, 2017-09) Murphy, Timothy S. ; Holzinger, Marcus J.

This work proposes a methodology for tasking of sensors to search an area of state space for a particular object, group of objects, or class of objects. This work creates a general unified mathematical framework for analyzing reacquisition, search, scheduling, and custody operations. In particular, this work looks at searching for unknown space object(s) with prior knowledge in the form of a set, which can be defined via an uncorrelated track, region of state space, or a variety of other methods. The follow-up tasking can occur from a variable location and time, which often requires searching a large region of the sky. This work analyzes the area of a search region over time to inform a time optimal search method. Simulation work looks at analyzing search regions relative to a particular sensor, and testing a tasking algorithm to search through the region. The tasking algorithm is also validated on a reacquisition problem with a telescope system at Georgia Tech.
Empirical Dynamic Data Driven Detection Tracking Using Detectionless and Traditional FiSSt Methods

(Georgia Institute of Technology, 2017-09) Virani, Shahzad ; Murphy, Timothy S. ; Holzinger, Marcus J. ; Jones, Brandon A.

Autonomous search and recovery of resident space object (RSO) tracks is crucial for decision makers in SSA. This paper leverages dynamic data driven approaches to improve methodologies used in real-time detection and tracking of RSOs with a low signal-to-noise ratio (SNR). Detected RSOs are assigned to be tracked using one of two simultaneously operating algorithms. The Gaussian Mixture Proability Hypothesis Density (GM-PHD) filter tracks all RSOs above a certain SNR threshold, while a Detectionless Multi-Bernoulli filter (D-MB) detects and tracks low SNR objects. The D-MB filter uses matched filtering for likelihood computation which is highly non-Gaussian for dim objects. Hence, the D-MB filter is particle based which leads to higher computational complexity. The primary idea proposed in this paper is to balance the computational efficiency of GM-PHD and high sensitivity of the D-MB likelihood computation by dynamically switching tracks between the two filters based on the SNR of the target; allowing for real-time detection and tracking. These algorithms are implemented and tested on real data of objects in the geostationary (GEO) belt using a wide field-of-view camera (18.2 degrees). A star tracking mount is used to inertially stare at the GEO belt and data are collected for 2 hours corresponding to RSOs being observed in 48.2 degrees of the GEO belt.
Uncued Low SNR Detection with Likelihood from Image Multi Bernoulli Filter

(Georgia Institute of Technology, 2016-09) Murphy, Timothy S. ; Holzinger, Marcus J.

Both SSA and SDA necessitate uncued, partially informed detection and orbit determination efforts for small space objects which often produce only low strength electro-optical signatures. General frame to frame detection and tracking of objects includes methods such as moving target indicator, multiple hypothesis testing, direct track-before-detect methods, and random finite set based multi-object tracking. This paper will apply the multi-Bernoulli filter to low signal-to-noise ratio (SNR), uncued detection of space objects for space domain awareness applications. The primary novel innovation in this paper is a detailed analysis of the existing state-of-the-art likelihood functions and a likelihood function, based on a binary hypothesis, previously proposed by the authors. The algorithm is tested on electro-optical imagery obtained from a variety of sensors at Georgia Tech, including the GT-SORT 0.5m Raven-class telescope, and a twenty degree field of view high frame rate CMOS sensor. In particular, a data set of an extended pass of the Hitomi Astro-H satellite approximately 3 days after loss of communication and potential break up is examined.
Direct Image-to-Likelihood for Track-Before-Detect Multi-Bernoulli Filter

(Georgia Institute of Technology, 2016-02) Murphy, Timothy S. ; Holzinger, Marcus J. ; Flewelling, Brien R.

This paper aims to apply the random finite set-based multi-Bernoilli filter to frame to- frame tracking of space objects observed in electro optical imagery for space domain awareness applications. First, this paper will review random finite set filters applied to frame to frame tracking and their applications to space. A new likelihood function for space based imagery will be presented, based on the matched filter. A more educated birth model will be proposed which better models potential SO using observer characteristics and object dynamics. Simulation results will explore the range of objects that can be tracked. The final algorithm is able to perform completely uncued detection down to a total object SNR of 5.6 and a per pixel SNR of 1.5. Promising but inconclusive results are shown for total object SNR of 3.35 and per pixel SNR of 0.7.
Hardware-in-the-Loop Comparison of Space Object Detection Tracking Methodologies

(Georgia Institute of Technology, 2016-02) Lee, Jared ; Zubair, Lubna ; Virani, Shahzad ; Murphy, Timothy S. ; Holzinger, Marcus J.

Space Domain Awareness relies on a network of surveillance architectures to catalog passive and active space object data. This paper analyzes the performance of several detection and tracking algorithms using different hardware-in-the-loop test configurations. Several image sources are tested, including simulated images generated with the Hipparcos and Space Object catalogs, and real images taken from a Nocturn XL camera. The Moving Target Indicator (MTI), Multiple Hypothesis Tracking, and Finite Set Statistics algorithms are tested, implemented, and compared in this study.
Spatio-Temporal Scale Space Analysis of Photometric Signals with Tracking Error

(Georgia Institute of Technology, 2015-09) Flewelling, Brien R. ; Murphy, Timothy S. ; Rhodes, Andrew P. ; Holzinger, Marcus J. ; Christian, John A.

This paper will investigate the application of Scale-Space Theory, specifically Curvature Scale Space, to 1-Dimensional light curve signals generated by reducing imagery sequences taken from simulated telescopes tasked in various modes. As an observed object with a variable light curve is viewed from a sensor achieving a perfect rate track mode, there is a trade between the time fidelity of the reconstructed signal and integration time required to make accurate detections. As the tracking error increases, a sensor in a step-stare con-ops for example may trade spatial samples for intensity information as a function of time. This is commonly seen in streak observations of tumbling resident space objects. The method presented here will demonstrate how consistent light curves with maximum time resolution can be generated from observation sequences with variable tracking error, and sensor integration times. Additionally, the sparse representation of these signals using Curvature Scale-Space feature images will be investigated as a means for rapid correlation of light-curves against a large database. The proposed rapid correlations could be used to identify variable operating modes of a known object, or to identify an object as a member of a database using a method dependent on the order of the number of salient features as opposed to the number of observations.
Orbit Determination for Partially Understood Object via Matched Filter Bank

(Georgia Institute of Technology, 2015-08) Murphy, Timothy S. ; Holzinger, Marcus J. ; Flewelling, Brien R.

With knowledge of a space object's orbit, the matched filter is an image processing technique which allows low signal-to-noise ratio objects to be detected. Many space situational awareness research efforts have looked at ways to characterize the probability density function of a partially understood space object. When prior knowledge is only constrained to a probability density function, many matched filter templates could be representative of the space object, necessitating a bank of matched filters. This paper develops the measurement dissimilarity metric which is then applied to partition a general prior set of orbits. A method for hypothesis testing the result of a matched filter for a space object is developed. Finally, a framework for orbit determination based on the matched filter result is developed. Simulation shows that the analytic results enable more efficient computation and a better framework for implementing matched filters.
Enabling Direct Feedback Between Initial Orbit Determination and Sensor Data Processing for Detection and Tracking of Space Objects

(Georgia Institute of Technology, 2015-04) Sease, Brad ; Murphy, Timothy S. ; Flewelling, Brien R. ; Holzinger, Marcus J. ; Black, Jonathan

This paper presents an automatic RSO detection and tracking scheme operating at the optical sensor system level. The software presented is a pipeline for processing ground or space-based imagery built from several sub-algorithms which processes raw or calibrated imagery, detects and discriminates non-star objects, and associates observations over time. An orbit determination routine uses an admissible region to start off an unscented particle filter. This preliminary orbit estimate allows prediction of the appearance of the object in the next frame. A matched filter uses this imagery to provide feedback to the initial detection and tracking process.
Particle and Matched Filtering Using Admissible Regions

(Georgia Institute of Technology, 2015-01) Murphy, Timothy S. ; Flewelling, Brien ; Holzinger, Marcus J.

The main result to be presented in this paper is a novel matched filter based on orbital mechanics. The matched filter is an image processing technique which allows low signal-to-noise ratio objects to be detected. By using previous orbital knowledge, the matched filter utility can be increased. First, the particle filter implementation will be discussed followed by the implementation of the matched filter. Then a pair of simulation results will be presented, showing the results from the particle filter and matched filter.