Gait Recognition

Gait Recognition Photo Credit: Clipart.com

Whether you've got a bounce in your stride or a swing in your step, the way you walk is more distinctive than you might realize. And it may contain important clues about what you're doing. That's why scientists are developing ways to identify and analyze walking patterns from a distance. You'll hear about one of them in this Science Update.


A sensor for skulk and swagger. I'm Bob Hirshon and this is Science Update.

You can tell a lot about people from the way they walk, according to Georgia Tech research engineer Bill Marshall. He and his colleagues are developing a radar system that can recognize and analyze a person's gait—whether it's a quick shuffle or a purposeful stride.


Essentially, in our system, we transmit radio frequency energy; it's reflected off the individual as they're walking toward the radar.

As the reflected waves shift, they're converted into an audio signature, which can then be analyzed and catalogued.

Marshall says walking patterns aren't necessarily unique to each person. But unlike fingerprints and voices, they can be analyzed without enrolling people in a database. What's more, his system works from up to five hundred feet away, regardless of visibility or what the target person is wearing. And that could provide useful information.


Instead of trying to decide if this is a particular individual, you might be trying to ask the question: is this individual walking with too short of a stride given their height? Or is this individual walking at too great of a speed?

He says clues like these could help spot people behaving suspiciously in sensitive areas, like airports, embassies, or military facilities. For the American Association for the Advancement of Science, I'm Bob Hirshon.

Making Sense of the Research

The science of identifying people by their physical characteristics is called biometrics. Biometrics is becoming an important field in science for several reasons, not the least of which is the heightened demand for security in a variety of situations.

Some kinds of biometric identification systems, like those that use patterns of blood vessels in the retina (the back of the eye) are very accurate. But generally, in order for a system like that to identify you, you have to voluntarily enroll yourself in its central database. Then, when you try to gain access to a facility or to classified information, the system takes a scan of your retina and matches it to that stored file to verify your identity.

That works great when you're trying to selectively identify a small group of people with special access privileges. But you can't positively identify anybody who isn't enrolled in the system; you can tell who they're not, but not who they are. Furthermore, you generally have to interact closely with these systems in order for them to check your identity (for example, walk up to a retinal scanner and push some buttons). Obviously, someone with malicious intentions would avoid that. So, if you want to watch out for suspicious or dangerous individuals from a distance, these systems won't help.

Gait recognition is one kind of biometric technology that can be used to monitor people without their cooperation. Some researchers are working on visually-based systems that use video cameras to analyze the movements of each body part—the knee, the foot, the shoulder, and so on. Marshall's team uses a radar-based system. As an individual walks toward the system, they're bombarded with invisible radio waves. Each individual's walking speed and style will make those waves bounce back a little differently. The result is a kind of composite signature that characterizes the overall feel of their walk.

This information couldn't pick out someone on the FBI's Most Wanted List if they wandered into an airport. But it could be used to spot people who are moving around in suspicious ways, which may include repetitive walking patterns (suggesting they're "casing out" a target) or movements that don't appear natural given their physicality. It could also be used in conjunction with other biometric systems to verify people's identities—and perhaps even weed out imposters.

Now try and answer these questions:

  1. What are biometric systems? Give examples.
  2. What are the advantages of the gait recognition system? What are the disadvantages?
  3. What kind of identification system might be best in the following situations?
    • Identifying employees at a military base
    • Watching for shoplifters in a store
    • Watching for terrorists outside a nuclear power plant
    • Maintaining security at an airport
  4. Suppose this kind of system were installed in a public place (like a busy subway station) to look out for suspicious activity. What kinds of complications might present themselves as the system was put into use? If you were in charge of the system, what steps could you take to minimize those complications?
  5. What ethical questions arise whenever you monitor and classify behavior as suspicious and non-suspicious?


For Educators

Georgia Tech's Human Identification at a Distance site features another kind of gait recognition research based on video imaging.

The University of Southampton, United Kingdom also has a gait recognition research page, with downloadable data from some of their experiments.

The Biometric Consortium is the United States Government's focal point for research, development, testing, evaluation, and application of biometric personal identification and verification technology.

Related Resources

Robotic Arm
6-12 | Audio
Mobile Plastic Fuel
6-12 | Audio
Driving Music
6-12 | Audio

Did you find this resource helpful?

Science Update Details

Grades Themes Project 2061 Benchmarks

Recent Science Updates

  • Arctic Greening

    30 years of satellite data confirms that the Arctic is greening.

  • Lego® Nerve Gas Detector

    Chemists design a mobile, affordable nerve gas detector out of Lego® bricks.

  • Shrimp-Inspired Navigation

    The mantis shrimp’s ability to see circular polarized light inspires an underwater GPS system.