Skip to main content

Soda machine serves as biometrics training ground

On the LCD screen, soda machine users can monitor face and fingerprint recognition in real time.
On the LCD screen, soda machine users can monitor face and fingerprint recognition in real time.

With a name like "SodaTron," it has to be high tech. Computer science and engineering (CSE) graduate students have programmed their student lounge vending machine to recognize faces and fingerprints, instantly linking users to cash stored in electronic accounts that are part of the CSE snack cooperative "Chez Bob."

Welcome to SodaVision, the latest in a series of student-run projects revolving around a conventional soda machine purchased by CSE professor Stefan Savage in 2005. "My hope was that students would come up with all kinds of crazy ideas and work together to make them real," says Savage.

Since then, CSE graduate students have torn apart and reassembled the machine's electrical nervous system, and added a computer, a touch screen display, a web cam, a fingerprint reader and a barcode scanner. Students have also designed and implemented the software to control the hardware. No mere prototype, SodaTron is used every day by the department's graduate students, faculty and staff.

Users simply walk up to the machine in the graduate student lounge in the Computer Science and Engineering building and position themselves squarely within the sights of the web cam mounted at eye level. The system must first recognize the presence of a face. Next, the software alters, cuts up and processes the image of the face captured by the web cam in preparation for recognition analysis.

The LCD screen attached to the soda machine— with its Star Trek inspired graphics—displays the progress of the facial recognition process, which takes a few seconds.

In an attempt to bypass day-to-day differences in an individual's appearance, the students wrote software that analyzes just the region around the eyes, explains Tom Duerig, the student leader of the collaborative project. Eyeglasses pose no problem to the system, and eyebrows can be in any position.

Students did the bulk of the programming, implementation and testing for SodaVision during the spring and summer of 2006.

SodaTron has become a platform for students to try out new research ideas. Ph.D. student Mikhail Afanasyev, who built the custom hardware to interface the machine to a computer, is hard at work building a robot to deliver soda door-to-door. Others are exploring how to add voice recognition, custom paper currencies and a 3-D user interface. "Its half laboratory and half playground," says Savage,"Each time I buy a soda I'm ready for a surprise.