AEM faculty spotlight:
Demoz Gebre-Egziabher
Calls to 911 are often unnerving, more so when the calls are made from a cell phone. When emergency calls are made from cell phones, it is difficult for authorities to precisely identify the position of the caller. Even though GPS will help with this problem, there are still challenges because in rural areas and in buildings and tunnels, GPS signals transmitted by cell phones often fail. Research conducted by AEM faculty member Demoz Gebre-Egziabher aims to change some of that.
Dr. Gebre-Egziabher, a McKnight Land Grant Professor at the University, is researching ways to make GPS systems more resistant to natural and artificial interference. He is relying on “sensor fusion,” which incorporates a variety of signals – GPS, TV, radio, and the like – to determine an incredibly accurate location from the data. In addition to implications in the world of emergency response, Gebre-Egziabher’s research could help mold safer aerospace systems like miniature aerial vehicles and autonomous automobiles.
What follows is a conversation with Prof. Gebre-Egziabher in which GPS, sensor fusion, and implications for emergency response are discussed in more detail.
![James](/people/faculty/bio/gebre/gebre.jpg)
Demoz Gebre-Egziabher
Why focus on improving GPS systems?
GPS is a wonderful system which has a performance that is superior to most, if not all, other navigation and timing systems. That is why people from all walks of life have become “addicted” to GPS. One of the most important uses of GPS is as an accurate timing standard. Most people are not aware of this, but GPS has become the “drum beat” to which timing systems like the ones used in communication and power networks are synchronized. The problem is, however, that GPS has some obvious weaknesses. For example, the casual user of GPS knows that it won’t work in tunnels or in deep urban canyons where there are buildings all around you. It is also very easy to interfere with its signal if you have malicious intent. This is a concern. For example, if there is a national emergency, there is so much that depends on GPS that recovery efforts may be difficult if the emergency also affected GPS.
We are trying to devise ways of making it robust. This means making it resistant to malicious interference or general failures. One of the ways we are doing this is by looking at GPS as just one signal in a system. We aid it by using many other signals, like from cell phones, other navigation systems, and the like. This is what is sometimes called the field of “sensor fusion.”
How would this sort of system work?
All this information fusion will be transparent to the users of GPS. A GPS unit that fuses all this information will start listening to not only GPS, but other information sources like TV signals, wheel speed sensors (in cars) and step counters (in personal navigation systems that, for example, can be used to guide the blind in a city environment). A unit will process whatever information it can listen to and try to combine it in such a way that it allows the unit to figure out where you are, any time or any place. The world is awash with all these signals that we do not use, but there is a lot of information in, say, TV signals that can be sued to aid GPS and figure out where you are.
What is the future of sensor fusion in aerospace?
A larger future challenge is creating very reliable sensor fusion. Let us say you’re trying to land an airplane on the runway in a heavy fog. Pilots can’t see anything, so they are using a navigation system to guide them down. The navigation system is going to be generating landing solutions based on several different sources. Now suppose one of the system is generating bogus information. That could end up affecting the final solution. This is not only a concern for aerospace applications. Consider the application of an automated vehicle on a highway or a guiding system for the blind that is based on combining information from many sensors. The risk of something dangerous happening to the user as a result of a failed or erroneous sensor is what we are trying to minimize. So one of the central questions in such sensor fusion problems is how to detect bad information and get rid of it in a timely manner.
Can interested students get involved in your research?
Absolutely. For example, the Nanosat project is a student satellite project I supervise. In this project, students design a satellite from the ground-up, is a classic example of sensor fusion. You have one satellite that is going to take information from various sources and combine it to do something useful.
Additionally, there is an opportunity to work with me and other faculty members in the AEM UAV group. In this group we are helping to build small airplanes that will be used in applications such as environmental monitoring or collection of traffic data. The students are building systems for these airplanes which fuse information from cameras, GPS and also computer graphics interfaces.
Last Modified: Thursday, 01-Nov-2007 11:28:46 CDT -- this is in International Standard Date and Time Notation