Pocket-sized GPS navigation systems can help pedestrians find their way anywhere on earth. Well, almost anywhere. Signals from GPS satellites cover the earth, but there are places they can't penetrate: in caves, in mountainous terrain, and in urban canyons—environments where soldiers and emergency responders need to go and quickly find their way out again.
Adding inertial navigation helps. Using accelerometers and gyroscopes, an inertial navigation system small enough to be mounted on—or even in—a shoe can track how far and in what direction a person has gone since the last GPS reading. Such systems by themselves, however, aren't nearly accurate enough. Accumulated integration errors and varying self-generated sensor output cause drift, making the equipment seem to register movement even when the user is standing still. That can cause surprisingly large position errors during even brief walks.
A paper published in the October IEEE Transactions on Microwave Theory and Techniques suggests an interesting solution.
GETTING THE DRIFT
Since drift occurs even when the navigation-equipped foot is stationary, measuring apparent motion when there actually is none yields correction data that can be applied to readings made from a moving foot—a technique called zero velocity update. And since a walker's feet are stationary each time he or she presses against the ground (except when slipping), drift calculations can be updated with every stride—provided the system knows when the instrumented foot is actually still.
For inertial navigation systems, that's not as simple as it sounds. The solution suggested by the paper's authors—IEEE members Chenming Zhou and Tamal Mukherjee, Student Member James Downey, and Fellow Daniel Stancil, then all with Carnegie Mellon University, in Pittsburgh—is apparent from its title: "A Low-Power Shoe-Embedded Radar for Aiding Pedestrian Inertial Navigation."
Unlike most radar systems designed to detect distant objects and their location and trajectory, this system needs only to detect when the user's heel is stationary relative to the ground or floor. For that, the radar need cover only distances of less than a meter.
"The radar computes the shoe's velocity relative to the ground, and only needs to determine zero velocity when the shoe and the ground are in contact, by which point the antenna within the shoe's heel is only a couple of centimeters from the ground," Mukherjee says. The radar has no need to scan an area; it measures only downward from the measuring shoe's heel. That helps keep the system's power consumption low enough for practical battery operation.
With an operating frequency of 6.7 gigahertz, the system's transmitting and receiving antennas are small enough to fit into the heel of a shoe. The system's electronics, a terrain relative-velocity-sensor op-amp, data acquisition board, and interface circuitry, all off-the-shelf components, have been mounted externally, but the authors anticipate size and power reductions with further hardware development.
Stancil says they have a prototype on which the entire radar and antennas fit on a circuit board which, at about 44 by 45 millimeters, fits within a typical heel. That does not include data acquisition and inertial navigation circuits, currently mounted on the shoe's surface and communicating via cable with a laptop computer; but in a future device they could all be embedded inside the shoe and communicate their correction signals wirelessly to a handheld navigation system. That could be done via Bluetooth or Wi-Fi, both at about 2.4 GHz, different enough from the prototype's 6 to 8 GHz to avoid interference.
"We have not looked into security problems, or how to make the unit resistant to jamming or tracking, but these are problems common to many wireless systems, and promising solutions do exist," Stancil says. He is now a professor and head of the department of electrical and computer engineering at North Carolina State University, in Raleigh.
"The performance of the radar system has been tested on different surfaces, and it worked on most, including concrete, grass, wood, sand, and carpet," says lead author Chenming, now at Disney Research in Pittsburgh.
Stancil adds there are some differences in reflection from the surfaces, but in all cases, the changes stop when the foot is stationary. Soft surfaces such as thick grass might be an exception, but most surfaces provide a clear, unambiguous indication of when the heel is stationary.
The initial impetus for the project came from the Defense Advanced Research Projects Agency, a wing of the U.S. Department of Defense.
"If you have a military mission someplace where soldiers might not be able to count on GPS, you want to be sure they can find their way back," Stancil says.
But soldiers are not the only potential users. "Our system could actually benefit all kinds of people who need localization information when GPS signals are not available, or reliable," Chenming says. They could be coal miners or first responders who go into collapsed buildings.
Cave explorers might find it useful, too, though Chenming points out that the initial models will be too expensive for recreational users.