May 10, 2001
By YUDHIJIT BHATTACHARJEE
ON a cold February afternoon this year, a wheelchair equipped with sensors and a computer rolled across the busy concourse of a railway station in Ulm, Germany, dodging passengers and piles of baggage. A group of engineers watched intently from one end of the concourse as a man stepped in front of the wheelchair. It stopped and turned right, stopped again and turned left as the man danced around, repeatedly blocking its path. Finally, a metallic voice from the wheelchair asked in polite German, "Could you please give way?"
The spectacle, captured on videotape, evoked cheers from the wheelchair's engineers, robotics experts from the University of Ulm and NASA's Jet Propulsion Laboratory. The self-navigating wheelchair, part of a class of smart machines that rely on advances in artificial intelligence and sensor technology, could give the most severely disabled a new degree of autonomy.
"Many handicapped people are incapable of even slight finger movements," said Dr. Erwin Prassler, an engineer at the University of Ulm's Research Institute for Applied Knowledge Processing. "They cannot use ordinary wheelchairs that have to be controlled by a joystick."
Dr. Prassler and his colleagues are designing the autonomous wheelchair, called MAid (for mobility aid), for people with such severe disabilities — a group that includes car crash survivors, polio victims and patients with multiple sclerosis. MAid's final version should be able to glide through any crowded environment, whether it is a Cairo bazaar or a New York subway station.
"All the user has to do," Dr. Prassler said, "is speak into the wheelchair's speech-recognition system — for example, ask the wheelchair to go 50 meters in a certain direction. The user could also indicate a target position on a touch panel that displays a map of the surroundings. This could be done either by touching the display or by voice commands."
The wheelchair then charts the best course. To judge the locations of nearby objects, a laser radar sends out light pulses and measures the time it takes for the reflections to be received. Other sensors keep track of the wheelchair's own position and speed.
By comparing successive radar measurements, the navigation system identifies stationary and moving objects in the vicinity. The comparison also tells how fast an object is moving and in what direction. An onboard computer figures out whether any of the moving objects are on a collision course with the wheelchair and computes a new path accordingly. All the steps in the navigation process are repeated several times each minute so the journey is smooth.
Dr. Prassler's group is working to upgrade MAid's intelligence so the wheelchair can respond to difficult situations, like being deliberately blocked by someone. The researchers are doing that by developing decision-making software, which they tested successfully at the railway concourse in February.
MAid is still some distance away from the market, but other sophisticated mobility aids for the less severely disabled are almost there. The closest is Johnson & Johnson's iBOT, a four-wheel chair that can climb stairs, ride over curbs and elevate the user to eye level with the walking world. It was developed by Dean Kamen, an inventor, best known for his latest project, code-named Ginger, which is rumored to be some kind of personal transportation device.
The iBOT wheelchair stands up on its rear wheels while moving up a staircase. An electronic gyroscope-assisted lift system raises the seat if the user wants to grab something off a high kitchen shelf. The chair can travel across rough terrain. Once it is approved by the Food and Drug Administration, which could take until the end of next year, the device is expected to sell for $20,000 to $25,000. Ordinary electric wheelchairs cost $5,000 to $10,000.
At the Royal Institute of Technology in Stockholm, Dr. Henrik Christensen and his colleagues have developed a module that can help wheelchair users navigate with better accuracy. The unit is a combination of sensors and a computing device. A wheelchair fitted with the unit can be steered safely around obstacles even by users with only partial control over the joystick. Dr. Christensen said the device could save wheelchair-bound people embarrassment.
"If the user wants to go through a doorway," Dr. Christensen said, "he or she has to only point the joystick in the desired direction."
A Swedish wheelchair company, Permobil, is working to commercialize the device. Dr. Christensen said the final product could cost anywhere between $800 and $1,500.
Engineers are also exploring ways to make existing wheelchairs more useful. The JWII from the Yamaha Corporation, for example, lets people operate manual wheelchairs without straining their arms. The unit provides just enough power to complement the user's propelling efforts. It is already available in Europe and is under review in the United States.
"Operating wheelchairs manually can put a lot of strain on the upper part of the body," said Dr. Rory A. Cooper, a University of Pittsburgh professor who is conducting studies using the Yamaha unit, "leading to degeneration of nerves and joints. This form of secondary disability can be prevented with the help of power-assist units."
For elderly patients who use wheelchairs and stay in their homes, a semiautomatic indoor navigation system being developed at Kanazawa University in Kanazawa, Japan, could reduce the need for care. Beacons on the ceiling would track the position of the wheelchair with ultrasonic pulses, and a control system would help the patient move from one room to another.
But for the most
severely disabled, nothing short of a fully autonomous system, like MAid,
will be able to provide real independence. The first MAid prototype may
be used at the Opera don Calabria rehabilitation center in Verona, Italy.
"For some of our patients," said Stefano Schena, director of the center,
"it could mean a new life."