2005 Bowman Scholars

Non-Orthogonal Iris Localization

The goal of this Bowman Scholar project is to isolate the iris in a non-orthogonal digital image of the human eye; that is, an image where the eye is not looking directly at the camera.  The iris (i.e., the colored part of the eye)  is a unique biometric identifier.  Much like a fingerprint, an iris’s pattern differs significantly between individuals (including identical twins), and analysis of these patterns can lead to accurate personnel identification.

Presently, iris recognition algorithms exist in commercial systems, and these systems are becoming more widespread in government and industry for logical security and access control.  Unfortunately, current systems require a cooperative user to interact with image acquisition devices.  The requirement for a cooperative user is based largely on the need for a “forward facing” image.  The iris recognition algorithms assume these images are normal to the sensing devices, and therefore search for circular patterns in the image.  Off-angle, or non-orthogonal, images of irises cannot currently be used for identification because the iris appears elliptical; the algorithms are incapable of isolating the iris in order to start the identification process.

The main goal of this research is to expand the functionality of iris recognition technology by developing a set of new algorithms to isolate the iris in a non-orthogonal state within a digital image.  Coding for the algorithms will be done primarily with MATLAB 7.0.  The algorithmic approach for successful development will be to first isolate the pupil, the dark homogeneous portion of the eye.  Once the pupil has been isolated, limbic boundaries (the outer edges of the iris) will be determined in multiple directions.  At no point do the algorithms assume that patterns are circular within the iris image.  An elliptical curve will then be fitted to these limbic boundary locations in order to create the “iris mask”.  The mask will contain almost all of the iris data which can subsequently be processed for identification.  The functionality of the algorithm will be tested using two separate databases: the CASIA (Chinese Academy of Science, Institute of Automation) and an in-house database collected at the United States Naval Academy.

With the ability to identify individuals in a non-ideal state, covert identification becomes feasible.  No longer does a user have to interact with an acquisition device and look directly at a sensor.  Instead, the sensor only needs to capture the iris at one of a wide-range of angles.  With these capabilities, covert iris recognition can move out of the movies and into reality, a necessary step for national security and our nation’s war on terrorism.

An Historical and Applied Aerodynamics Study of the Wright Brothers' Wind Tunnel Test Program and Application to Successful Manned Flight

This project is a study of the Wright Brothers’ wind tunnel research:  what they studied, what they learned, how they applied their knowledge quantitatively and qualitatively, and how the wind tunnel contributed to their success.  The Wrights were brilliant engineers, but made several critical design decisions based on partial information, imprecise results, and even complete misunderstanding.  This leads to two possible conclusions: i) they were extremely lucky in making the right decisions at crucial points, or ii) they were extremely astute in recognizing the limitations of their experiments and the shortcomings of their results, and compensated accordingly.

The greatest contributor to the Wrights’ success in was their wind tunnel research.  From this research they acquired the aerodynamic understanding used to design their Wright Flyer.  Considering the importance of their wind tunnel research, the topic has been studied very little.  To date, an in-depth analysis of the Wrights’ wind tunnel experimentation has not been conducted.  Specifically, there is no full understanding of the Wrights’ wind tunnel capabilities and limitations, and, thus, no understanding of how they dealt with those capabilities and limitations.

Such a full understanding of the Wrights’ wind tunnel research and methods requires both historical and applied analysis of the Wrights’ experimentation.  The historical research involved studying primary sources at the Smithsonian Air and Space Museum under the guidance of Dr. John Anderson and Dr. Peter Jakab, as well as independent research at the Library of Congress.  The purpose of the historical research was to analyze that information to which the Wrights had access, to see what information they accepted and which they ignored, and to track the development of their aeronautical education.

The bulk of the research involves studying the Wrights’ wind tunnel and experimental methods.  By constructing a replica tunnel with flow matching that of the Wrights’ tunnel, a detailed analysis of the tunnel’s flow characteristics can be conducted, as well as an analysis of how those characteristics affect the results generated by the tunnel.  This research, combined with selective numerical analysis of the Wrights’ airfoils and wings, provides an understanding of how the wind tunnel itself, and the way in which the Wrights used the tunnel, affected the aerodynamic knowledge they gained from their experimentation.

Combining applied and historical research presents a clear picture of the Wright’s aeronautical background, growth, and final understanding of aerodynamics which ultimately led to their success with the Wright Flyer.    

Study of the Beta Response Characteristics of the DT-702 Thermoluminescent Dosimeter

The Navy is currently in the process of employing a new generation four element thermoluminescent dosimeter, the DT-702. The focus of this study is to improve radiation dosimetry in the United States Navy by modifying the existing the DT-702 dose algorithm so that it can more accurately calculate beta exposures in radiations fields that contain a mixture of gammas, neutrons, and betas. Experimental data was collected at NIST in which DT-702 dosimeters were irradiated using both Sr-90 and Cs-137 beta sources. This data will be analyzed with the Monte Carlo radiation transport code, MCNP, in order to better understand the DT-702’s response in mixed radiation fields. The results of this analysis will be used to develop decision tree coefficients, which can then be integrated into the existing dose algorithm for the DT-702.

Property and Structure Evaluation as a Function of Processing Parameters: Large HY-80 Steel Castings for a U.S. Navy Submarine

This Bowman Scholar project involves the investigation of the processing parameters of HY-80 steel castings.  The properties in metals are not only a function of the composition. Processing techniques can significantly alter how a metal performs and can ultimately determine whether a material will fail.  It is imperative that metals be processed correctly so that components do not fail before their expected lifetime.  Preliminary studies indicate that the HY-80 Bridge Access Trunk casting failed while in service due to improper processing.  In this effort, samples taken from the failure location will be evaluated.  Heat treating furnaces and quenching apparatus at USNA will be utilized to austenitize sections of casting and then subject each to different quench and temper conditions. Mechanical properties and microstructural characteristics will be determined.  Tensile, impact, and hardness measurements, and electron and optical microscopy will all be used to evaluate the performance of high strength steel. The window(s) of favorable processing parameters defined by heat treatment temperature and cooling rate will be clarified based on the requirements set by the U.S. Navy for large HY-80 castings.

Developing a Safe and Efficient Surface Maneuvering Package for Submarines

The purpose of this project is to accurately model, observe, and measure submarine surface maneuvering characteristics in order to enhance the surface-operating envelope.  While encountering head-on seas, submarines have a tendency to plunge unexpectedly, creating a potentially dangerous situation for the crew topside and the equipment exposed to salt water.  In order to counter the tendency to plunge, submarines often run with a stern plane angle that will assist in keeping the bow from plunging.  However, any increased stern plane angle while maneuvering on the surface will create an added value of resistance.

While trimming the submarine on the surface is common practice to avoid plunging, no study has been completed that would suggest both an efficient and effective stern plane angle.  This project will vary the stern plane angle at a variety of velocity and sea state conditions.  Measurements will include resistance, pitch, heave, wave frequency, velocity, and stern plane angle.  Froude scaling enables a geometrically similar model to be built that will accurately simulate the full-scale responses of a surfaced submarine.  Results from these tests can then be applied to fleet platforms in order to aid in safe surfaced maneuvering procedures.

In addition to testing geometrically similar stern planes, higher aspect ratio stern planes will be placed on the model and tested.  The high aspect ratio stern planes have potential to reduce drag, thus increasing the efficiency.  While today’s submarines have relatively low aspect ratio stern planes, future design considerations may include higher aspect ratio control surfaces in order to improve both submerged and surfaced maneuvering. 

Evaluation of a Composite and Metal Hybrid Co-Cured Joint

This Bowman Scholar project involves the investigation of a novel hybrid composite joint in which a glass-reinforced composite material transitions into metal without the use of mechanical fasteners or adhesives.  The joint is fabricated by resin infusion and co-curing, meaning that the metal fibers overlapping the composite glass fibers are injected with resin simultaneously and then harden to form the joint.  The joint is the point of focus, as previous research has shown that the joint is much weaker than the two materials being joined.  The goals of this project include:  fabricate various joints, conduct tests to observe joint behavior, quantitatively describe behavior of different joints, and optimize the joint to provide maximum strength.  Several joints have been analyzed including symmetric and asymmetric stepped lap joints of different step lengths.  Analysis has shown that the strength of the hybrid joint is directly related to the step length and joint configuration.  There are numerous Naval applications for this material.  Currently, the only means of joining a composite material to a metallic material is by means of either a bolt or an adhesive.  Both methods of joining have inherent negative effects.  However, this technology would eliminate these problems, thus providing a much stronger transition.  Any design that may require a reduction in weight without a reduction in strength could benefit from this technology.  One area of interest for this technology is for the future Littoral Combat Ship (LCS).  The LCS is designed to operate in the littoral regions and operate at speeds in excess of fifty knots.  Therefore, the design must incorporate a shallow draft and a lighter displacement.  Alternatives to conventional materials must be used.  A hull made of this hybrid material would not only save on weight but would also allow for weapons systems and other systems to be easily joined to the hull.  Another possible application would be a hybrid shaft.  A metal shaft with the center span being of a composite material would greatly reduce the weight of the shaft.  This also would make shaft production easier since a large metal shaft would no longer need to be forged.

Investigation of Dielectric Charging in MEMS Capacitive Shunt Switches

Microelectromechanical Systems (MEMS) are microscopic systems that use electrostatic forces to do mechanical work.  Specifically, MEMS capacitive shunt switches use an electrostatic force to switch on or off a high-frequency electrical signal.  Such switches use very little power and offer better performance than current semiconductor-based microwave switches, and facilitate the development of low-power electrically-steerable radar. The MEMS capacitive shunt switch consists of a conductor, which carries the high frequency electromagnetic signal.  This conductor is surrounded on both sides by, but isolated from, a large conductor that is electrically grounded.  Another conductor, called the bridge, is suspended over the center conductor and connects both grounded conductors.  Between this bridge and the center conductor, there is a non-conductive layer called the dielectric layer.  When large DC voltages (20-40 V) are applied between the signal line and the ground plane, the bridge is brought into contact with the signal line, creating a short circuit for the high frequency signal that prevents transmission.  The dielectric layer prevents direct current (DC) conduction between the center conductor to the bridge.

One problem with capacitive shunt switches is a limited lifetime because of the build-up of charges in the dielectric layer, which degrade the performance of the device.  To better understand how to make MEMS capacitive shunt switches more reliable, the factors that affect dielectric charging must be known.  To measure the amount of charge built up in the dielectric, Midshipman Schultz will be implementing a non-contact measurement used by Dr. Reid, of the Air Force Research Lab.  The method is non-contact because it measures the dielectric charging without causing the bridge and the center conductor to come in direct contact.  By not allowing the center conductor and the bridge to come into contact, the measurement is not affected by new charge that would otherwise be introduced into the dielectric.  The measurement system will then be used to quantify the effects on dielectric charging of bias signal voltage and wave shape, heat and ionized air, and mechanical shock.

Temperature Characterization System and Communication Interface for Microelectromechanical Thermal Switches for the MidSTAR Satellite Program

This Bowman Scholar project deals with developing a platform for testing a microelectromechanical (MEM)thermal control system onboard the Naval Academy satellite MIDSTAR.  The platform must provide control and assessment of device operation, as well as communication with the ground station via the satellite communication system, referred to as the Command and Data Handling (C&DH) unit.  The satellite is set to launch in September of 2006.  This project will produce two deliverable items to the MidSTAR team.  One box consists of the electronic communications unit and control devices, while the second holds the temperature control device mounted with heaters and temperature sensors.

The two thermal switches serve as the experimental platform.  The first is an actively controlled electrostatic device that triggers with an applied voltage.   This switch is comprised of a foil suspended above a substrate base by polymer posts.  In the off state, the void between the foil and the base of the switch provides thermal isolation.  With an applied voltage, the foil bows downward to contact the base allowing the device to transfer heat.  The second device is comprised of a passive switch connected to a radiator.  At a certain threshold temperature, the device will activate and close the switch, connecting the body to the radiator.  In this state, the device serves to emit heat from the body of the device.

The communication system will interpret instructions from the C&DH unit to carry out experiment instructions prescribed for each orbit base upon the available power requirements.  It will also respond to inquiries by transmitting the collected data and reporting the status of the components.  The device collects data by digitizing temperature readings and stores them in the available flash memory inside the control unit.  These goals of data acquisition and communication are accomplished by programming a PIC16F874A Microcontroller.  With modularity being a key issue, the instruction set must be flexible enough to provide a wide range of operational capabilities to function within the power budget of the satellite.

The test for device effectiveness relies upon cycling the integrated heaters while operating the device.  If the MEMS functions as expected, there will be a discernable temperature difference between the active and inactive states of the thermal switches.

Investigation of a Chill Water Flow Rate, Surface Area and Heat Input for a Condenser System

The goal of this project is to determine an ideal number of cooling tubes to use in a condenser and steam generator system for a specified heat input from the steam cycle and flow rate from the chill water.  Using a condenser and steam generator designed and built during a mechanical engineering capstone course last year, the first step will be to hook up a water chiller to provide a heat path out of the system.  Second, the power source will be installed in the steam generator.  After checking the entire system for air leaks, experimentation can begin.  To do this, several characteristics of the system will act as variables.  In one test, the heat input will be fixed, while the number of chill water tubes and the Reynold’s number (the chill water flow rate) through the system are allowed to vary.  In the second test, the flow rate of the chilled water will be fixed, while heat input and number of chilled water tubes are varied.  Throughout both tests, the effects upon the saturation temperature, sub-cooling temperature, and overall heat transfer coefficient will be observed.

Advanced Economic Analysis as Applied to Issues of Economic Development

The goal of this Bowman Scholar project is to utilize advanced quantitative/analytic economic tools to evaluate the effectiveness of bilateral and multilateral trade agreements in Sub-Saharan Africa at achieving, through increased trade, the goal of sustained economic development. Sub-Saharan Africa is plagued, more than any other developing region, by staggering levels of poverty, lack of education, and lack of access to resources such as clean water. Much of development organizations’ efforts and research over the last fifty years have been focused on the region with varying, and sometimes disappointing, levels of success. The course will utilize econometric tools and real-world data to rate the effectiveness of various measures such as bilateral trade agreements at increasing trade volumes in the past, to explain Africa’s current economic situation, and to predict where poverty reduction and sustained development are headed in the future. Spatial analysis, gravity models, and simultaneous equations will primarily be utilized in order to determine which approaches have been most effective. By evaluating what methods have or currently show promise in Africa as well as evaluating successes and failures on a case-by-case basis, the course will ultimately facilitate judgment as to which measures are wastes of development resources, and which, such as cross-border market integration, may hold more promise for Africa’s poor. The mathematical tools utilized in this research, built on statistics and algebraic optimization, form a strong basis for analytic research in a variety of fields beyond economics, such as operations research, force planning, and public policy research.

Enhancing High Level Architecture Through Application of Contributions from the Object-oriented Method for Interoperability

The goal of this Bowman scholar project is to determine how High Level Architecture can be enhanced through concepts and ideas used in the Object-Oriented Method of Interoperability. Creating interoperability among heterogeneous systems enhances our military’s war-fighting capabilities. Differences in the hardware, languages, and the way systems represent data make interoperability hard to achieve. The Object-Oriented Method for Interoperability is one method used to resolve differences and therefore allow interoperability among heterogeneous systems. The OOMI Integrated Development Environment (OOMI IDE) constructs a Federation Interoperability Object Model (FIOM) prior to runtime that models information shared among the systems. At runtime the OOMI uses the information provided by the FIOM to resolve differences in information, therefore allowing systems to interoperate.

The "High Level Architecture" is a general-purpose software architecture currently used to allow interoperability among simulations. The HLA sets out guidelines that allow the simulations to run together, but it does not provide a method for identifying or resolving differences among corresponding objects modeled by the simulations. The OOMI includes these capabilities and incorporating them into the HLA would enhance its capabilities towards achieving component system interoperability.

Comparison of Metrics for Software Safety Prediction in Embedded Systems

The objective of this project is to analyze current metrics to determine their predictive role in preventing software from entering a hazardous state, an area known as software safety.  Such predictors, as well as analysis tools such as fault trees, input distributions, weakest preconditions, and fault injection, are beneficial for determining what level of verification is necessary to ensure the safety of the system.   Analysis of such tools and predictors plays an important role in deciding if a safety critical software component should be verified by formal methods or whether testing is sufficient to ensure safety.  Approaches such as fault injection use non-traditional testing techniques that examine how a software system reacts when known faulty conditions are forced to occur, and can be focused on the ratios of the number of seeded faults found by the various tools and metrics.

Expected results of this project include a determination of the effectiveness of metrics in safety applications, the feasibility of methods using fault trees and fault injection, and the general predictability of safety in software. As a case study, the flight software of the Naval Academy’s MidSTAR satellite will be tested and verified with respect to safety. Analyzing the effectiveness of the metrics in predicting safety software will demonstrate which, or whether, metrics can be considered a recommended tool in evaluating software safety. Indications of how predictable software safety is in general are anticipated based on the MidSTAR case study. The feasibility of using fault trees and fault injection as a predictor could present the basis for a new tool to use in testing software safety.

Classical and Quantum Singularities in Cylindrically Symmetric Spacetime

As the sciences of mathematics and physics progress, it is becoming more desirable to unify the different forces that describe the elements of nature within our universe.  Classical physics seems to accurately describe the properties of particles and bodies of mass that we commonly observe; however, this theory begins to fall apart when applied on the microscopic scale of atoms and electrons.  In the latter instance one needs a different theory, one commonly known as quantum mechanics.  It is the bridge between classical and quantum properties that has yet to be understood, much less perfected.  The goal of this Bowman Scholar Project is to contribute, albeit in a small way, to these studies.

Einstein’s theory of general relativity describes the behavior of classical particles in a gravitational field.  In general relativity, gravity is modeled by the curvature of abstract surfaces called spacetimes. In many cases, such spacetimes contain points that cannot be defined and particle behavior at or near these points is difficult to predict.  These points, widely known as singularities, are of distinct interest to the developing study of general relativity and its interrelationship with quantum mechanics.  Understanding the similarities and differences between the paths taken by quantum particles versus classical particles near a singularity can be revolutionary in unifying these two seemingly diverse theories.

Specifically, this project is a study of a class of spacetimes whose symmetry properties make them amenable to mathematical analysis; these spacetimes have the symmetry of an ordinary cylinder. One such spacetime is called a “wiggly cosmic string”.

It describes a predicted, but not yet observed, astronomical object. Its spacetime has both classical and quantum singularities, but neither have yet been analyzed mathematically. This project will examine these singularities and describe classical and quantum particle behavior near these singularities, serving as a small step in the larger search for a quantum theory of gravitation.

The study of general relativity is practical due to its important daily functions apparent in Global Positioning System services.  Currently, this is the main naval application of general relativity as it aids in ship navigation as well as ordinance delivery and accuracy.  However, with the advancement of the knowledge of quantum and classical principles, the unexplored opportunities and applications are endless.  For example, space technology and its importance to the United States military are critical reasons to study the development and unification of classical and quantum theories.

Induced Segmentation: The Key to Imitation Learning

To develop a system that enables a robot to learn through demonstration and imitation, a program must be created to analyze the stream of observed input data. In a robotic system, the basic unit of a plan may consist of many low level actions. Therefore, in an imitation learning scenario, the first step is to segment the stream of input data describing the demonstrator’s actions into chunks corresponding to the basic planning unit. Currently no such software exists.

Methods that segment the input data stream have already been attempted. In one case, the demonstrating robot sends the learning robot a signal each time a step is completed. This process is accurate; however, it requires the cooperation of the demonstrator, which may not always be available. Another method relies on the statistical properties of continuous and repetitive actions to induce the points of discontinuity in the stream. This method has not been fully explored and it is where the focus of this research project will reside.

Because this area is relatively unexplored, the analysis will begin with the input data streams. These streams consist of vectors of numbers. The overall goal is to identify the patterns and develop a probability equation that will be able to break down the segments into simple steps. Once the algorithm is developed, it will be implemented in a functional programming language. The results will be analyzed to determine if the algorithm was able to locate the segments correctly. This process will be repeated until the algorithm works and the research is completed. Then implementation of the algorithm into a robotics system will begin. Through this repetitive process, it is hoped that a solution to the segmentation problem will be identified.