15 research outputs found
First UND doctorate in computer science
No full textFirst UND doctorate in computer science
Bismarck native Kirk Ogaard is using his know-how to mine flight data for the Army\u27s Aberdeen Test Center.
You hear them flying overhead every day—they\u27re the aircraft that University of North Dakota aviation majors use to learn their craft. UND\u27s aviation program makes the Grand Forks Airport one of the busiest in the country in terms of takeoffs and landings.
But for Kirk Ogaard, there\u27s a very different kind of business associated with those aircraft: he mines flight data gathered directly from devices aboard.
Ogaard, originally from Bismarck, recently earned UND\u27s first Ph.D. in computer science, opening the path for several others behind him who\u27re also enrolled in the program. Ogaard also got his bachelor\u27s and master\u27s degrees in computer science from UND, known for its prowess as a center of learning in computational science.
For my Ph.D. dissertation, I wrote a program—a software package—to mine the data that we collected from airplanes used to train UND aviation students, said Ogaard. His successful Ph.D. completion won Ogaard a spot in a coveted one-to five-year post-doctoral program with the U.S. Army Aberdeen Test Center in Maryland. That\u27s one of nine such centers that support the Developmental Test Command, the Army\u27s premier organization for developmental testing of weapons and equipment.
At Aberdeen, I\u27ll be doing stuff similar to what I did for my Ph.D.—data mining and probably some visualization, said Ogaard, who plans to go into full-time research once he\u27s done with his post-doc.
I got help from Jim Higgins, a former captain with American Eagle Airline, who now teaches in UND\u27s aviation program, Ogaard said. He organized the system that collected all the data direct from the aircraft—such as global positioning system information—and offloaded it into a computer at the completion of each flight.
The challenge, he says, is that once you collect and mine data, there\u27s more than straight analysis.
To make the data analysis useful you need to be able to draw useful conclusions from it, Ogaard said. The real problem, then, is interpreting those results. Visualization can help you do that, whether you convert the answers into some sort of chart or other graphic—in other words, it helps you understand what you\u27ve found in the data.
You can use off-the-shelf software to create the graphics or you can write your own visualization software. Ogaard wrote his own.
What\u27s UND doing with Ogaard\u27s aviation data mining results?
Applicability is the key—the University can use it to look at the kinds of maneuvers that students perform, see which maneuvers are most frequent, Ogaard said. I think the most useful thing for the University is methodology I developed for analyzing and extracting value from the data.
Advice for the next generation of Ph.D.\u27s?
I would say most important thing, be persistent, keep working at, don\u27t get frustrated, Ogaard said. He completed his PhD in three and a half years after completing the two year\u27s master\u27s program. About UND PhD program in computer science
The Department of Computer Science offers graduate study leading to the Doctor of Philosophy in Scientific Computing (emphasizing the development of software, the science, and the technology required to support computational science and simulation based science and engineering). The department is a part of the John D. Odegard School of Aerospace Sciences, which provides unique opportunities for research by faculty and graduate students. There is especially strong interest within the department in the areas of artificial intelligence, compiler design, database, networks, operating systems, graphics, simulation, software engineering, and theoretical computer science.
Juan Pedraza
Writer and Editor, University Relation
Mining Aircraft Telemetry Data With Evolutionary Algorithms
No full textThe Ganged Phased Array Radar - Risk Mitigation System (GPAR-RMS) was a
mobile ground-based sense-and-avoid system for Unmanned Aircraft System (UAS)
operations developed by the University of North Dakota. GPAR-RMS detected proximate
aircraft with various sensor systems, including a 2D radar and an Automatic Dependent
Surveillance - Broadcast (ADS-B) receiver. Information about those aircraft was then
displayed to UAS operators via visualization software developed by the University of
North Dakota. The Risk Mitigation (RM) subsystem for GPAR-RMS was designed to
estimate the current risk of midair collision, between the Unmanned Aircraft (UA) and a
General Aviation (GA) aircraft flying under Visual Flight Rules (VFR) in the surrounding
airspace, for UAS operations in Class E airspace (i.e. below 18,000 feet MSL). However,
accurate probabilistic models for the behavior of pilots of GA aircraft flying under VFR
in Class E airspace were needed before the RM subsystem could be implemented.
In this dissertation the author presents the results of data mining an aircraft
telemetry data set from a consecutive nine month period in 2011. This aircraft telemetry
data set consisted of Flight Data Monitoring (FDM) data obtained from Garmin G1000
devices onboard every Cessna 172 in the University of North Dakota\u27s training fleet.
Data from aircraft which were potentially within the controlled airspace surrounding
controlled airports were excluded. Also, GA aircraft in the FDM data flying in Class E
airspace were assumed to be flying under VFR, which is usually a valid assumption.
Complex subpaths were discovered from the aircraft telemetry data set using a novel
application of an ant colony algorithm. Then, probabilistic models were data mined from
those subpaths using extensions of the Genetic K-Means (GKA) and Expectation-
Maximization (EM) algorithms.
The results obtained from the subpath discovery and data mining suggest a pilot
flying a GA aircraft near to an uncontrolled airport will perform different maneuvers than
a pilot flying a GA aircraft far from an uncontrolled airport, irrespective of the altitude of
the GA aircraft. However, since only aircraft telemetry data from the University of North
Dakota\u27s training fleet were data mined, these results are not likely to be applicable to GA
aircraft operating in a non-training environment
