Playing It by Ear

Perspectives: Research and Creative Activities at SIUC, Fall 2007

:: research survey ::

Playing It By Ear

An SIUC researcher's work on human-machine interfaces might give soldiers and people with profound disabilities more control over their activities and environment.

Lalit Gupta, professor of electrical and computer engineering, is a leader in harnessing signals generated by the human body and using them to control mechanical devices.

The body and brain constantly emit electrical impulses or other signals relative to whatever a person is doing, hearing, or seeing at any given time, Gupta says. Researchers can detect these signals through a variety of methods, including placing electrodes on a subject's head and recording the electrical activity.

The problem for researchers is tuning out the extraneous "noise" and identifying the specific signals they're looking for.

Gupta's work focuses on signal pattern recognition, or discerning the specific voluntary control signals generated by the body or by the brain in response to external audio (tones) or video (pictures) stimuli. Once he detects these patterns he can devise ways for the subject to control a device such as a computer or wheelchair by using those specific signals.

Recently, a company called Think-a-Move began marketing a wheelchair based on technology Gupta helped perfect. The wheelchair is controlled by the changes in air pressure inside a subject's Eustachian tube that occur when the person moves his or her tongue to different places inside the mouth.

Think-a-Move engineers had suspected that the tongue movements caused changes in the pressure, but needed a researcher to find a way to correlate the pressure changes with the different tongue movements. The company approached Gupta in 2003.

Gupta developed a strategy to filter the signals and to automatically detect, segment, and classify the ear pressure signals associated with moving one's tongue to the left, right, up, or down. SIUC students served as test subjects as Gupta determined the specific air pressure signatures associated with the tongue movements.

The system uses a small hearing-aid–like device that is actually a microphone. Once placed in the ear canal, it can detect the subtle air pressure changes associated with certain tongue movements and send them via a wire to a control unit hooked to the wheelchair.

"We were able to verify this approach works and got to where we could recognize these signals 97 percent of the time," Gupta says. The signals are largely the same for everyone, but the machine can be fine-tuned to the individual using it.

A person can control the wheelchair by moving his or her tongue to different positions. Where a person is capable of more complex tongue movements, such as flicking the tongue up twice, the system can respond to as many as 20 different commands. International media and scientific magazines reported on the technological breakthrough earlier this year.

Gupta also sees other potential applications for such devices and technology, such as devising communication methods for disabled persons who are fully aware but unable to communicate.

"It would help in very simple, binary conversations," he says. "A person would be able to answer yes or no, or communicate other ideas. There wouldn't be a large market for this, but for the people who need it, it would be very valuable."

The U.S. Army also is interested in a related technology as a means of hands-free control for vehicles. Gupta's research has shown that a similar device inside the Eustachian tube can discern spoken words. Because the device is inside the ear canal, it is shielded from the high ambient noise of the battlefield or heavy vehicles and thus is easier to use for voice command driving.

—by Tim Crosby