Engineers can build autonomous vehicles capable of cruising city streets without the aid of a human driver, as demonstrated two years ago in the DARPA Urban Challenge. A team of researchers is now looking to translate that success to the medical field by building so-called "smart wheelchairs" with artificial intelligence that uses lasers, sensors and mapping software to operate and navigate powered chairs for riders who cannot do so on their own.

With help from a five-year, $480,000 National Science Foundation grant received in June, a team of researchers led by John Spletzer, an associate professor of computer science and engineering at Lehigh University in Bethlehem, Pa., has developed a prototype chair designed specifically for negotiating sidewalks, parking lots and other outdoor areas.

As with other smart wheelchairs designed in the past, Spletzer's device uses a light detection and ranging (LiDAR) system to detect trees, poles, parking meters, corners and other real-world obstacles. A key difference is that this chair will cross-reference the maps it makes of its surroundings using LiDAR and other sensors with 3-D maps that Spletzer and his team create and load into its memory. To complement the chair's ability to recognize and avoid stationary obstacles, the researchers are also planning to write software that will help the chair predict and avoid moving obstacles such as pedestrians and cars.

"My work aims to push the envelope in wheelchair autonomy," Spletzer says. "It will not be limited to structured indoor environments. Instead, it investigates the much more difficult problem of autonomous operations in unstructured environments outdoors."

The prototype that Spletzer and his colleagues have built thus far can trace its lineage back to two of Spletzer's previous research endeavors. One is the 2007 DARPA Urban Challenge for robotic vehicles, where Spletzer was part of the team that produced "Little Ben," a Toyota Prius fitted with laser and camera sensors. Little Ben was one of six cars (out of a field of 89 vehicles) that autonomously completed the 92-kilometer course, designed to mimic the challenges of city driving.

Spletzer's work with Philadelphia's Freedom Sciences, LLC, likewise contributed to his smart wheelchair prototype. Freedom Sciences received U.S. Food and Drug Administration (FDA) approval in 2008 to sell its Automated Transport and Retrieval System (ATRS), which lets wheelchair users get into and out of their vehicles as well as stow and retrieve their motorized wheelchairs. ATRS works as such: after getting into the driver's seat, the driver uses a remote control to guide the wheelchair to a lift at the back of his car. Once there, the wheelchair uses a laser guidance system to dock itself on the lift platform, which is then retracted into the car. "The wheelchair only goes a couple of feet on its own," Spletzer says, "but it allows the driver to be independent."

Spletzer and his team created the maps for their prototype smart chair with Little Ben's help. They drove the modified Prius around South Bethlehem and the Stabler Arena parking lot on Lehigh's Goodman Campus to create 3-D maps later downloaded to their wheelchair's control system. (Although this sounds work-intensive, it is not unlike Google's approach to creating its Street View maps.) In test runs, the researchers have been able to get the chair to autonomously travel about one kilometer and arrive at its destination while tracking its position within an accuracy of 20 centimeters. Spletzer adds that they could not have achieved this level of accuracy using the Global Positioning System (GPS), which tends to be off by a meter or more. Not a big problem for drivers, but a huge issue for a wheelchair.

A key aspect of Spletzer's chair will be making it affordable. He points out that Freedom Sciences is able to sell its ATRS because that product's cost is comparable with that of purchasing a van or SUV and revamping it to allow a person to drive while seated in the wheelchair, as most wheelchair drivers now do. To get beyond the prototype phase with their smart wheelchair, Spletzer and his team will have to figure out how to get it to move autonomously without having to rely on the expensive LiDAR system that Little Ben used but rather on a combination of lower cost LiDAR and sonar sensors.

Of course, Spletzer and his colleagues are not the only ones who have come up with the idea of a smart wheelchair. A paper published in the December 1999 IEEE Transactions on Neural Systems and Rehabilitation Engineering describes a NavChair Assistive Wheelchair Navigation System (pdf) developed at the time to reduce the "cognitive and physical requirements of operating a power wheelchair for people with wide ranging impairments that limit their access to powered mobility." The NavChair was based on a commercial wheelchair system with the addition of a DOS-based computer system, ultrasonic sensors and an interface module interposed between the joystick and power unit .

Two more recent projects share nearly the same name: The Robochair project at the University of the Basque Country's Laboratory of Human–Computer Interaction for Special Needs in Spain is creating an autonomous navigation system that can be added to commercially available powered wheelchairs. In 2007 researchers at the University of Essex in England pursued a similar project called RoboChair (pdf).

Yet, whereas others have demonstrated that smart wheelchairs can work, they have yet to make them affordable for the large population of users (2.7 million people age 15 and older in the U.S. alone). "The ultimate goal is to develop a smart-chair system capable of unprecedented levels of autonomy while still being commercially viable," Spletzer says.

View a slide show featuring images of the prototype chair, a LiDAR map and Little Ben