With the price of guide dogs running into the tens of thousands of dollars, the white cane is still the most commonly used aid for the visually impaired. That venerable tool is the focus of research by Johann Borenstein and Iwan Ulrich, mechanical engineers at the University of Michigan. They have developed a cane that uses sonar to steer the cane’s user around obstacles.
The sonar cane looks a bit like a traditional white cane but has a semicircular black box on two wheels attached to its bottom end. As a person pushes the cane along, it transmits ultrasonic pulses that bounce off obstacles up to six feet ahead and reflect back to sensors at the base of the cane. The cane then automatically turns its wheels to steer around the obstacle. The person feels the movement and follows along, also sidestepping the object. It’s so intuitive to follow that within two minutes a person can be walking at about normal speed, says Borenstein.
The cane, which weighs eight pounds, contains steering motors for the two wheels, a small onboard computer to control the steering, and an array of ultrasonic sensors: eight look ahead, covering the forward 180 degrees; in a future model, one will look up for higher obstacles. The person sets the initial direction of travel with a thumb-activated joystick on the cane’s handle. When the cane detects an obstacle, the computer changes the wheel positions to avoid it. Once around the object, the cane returns to the original direction.
Borenstein and Ulrich must still solve a few problems. Some very smooth surfaces, such as a glass door, can reflect the ultrasound pulses away from the sensors, so they don’t detect the obstacle in time. And tables, for example, can exceed the reach of the sensors, making them hard to detect. The biggest problem is getting the cane to distinguish concrete from grass so that it can follow the meanderings of a sidewalk without veering onto a lawn. As with a non-sonar cane, the user can feel the difference, but Borenstein says the sonar cane itself will be able to handle this situation in a few years, for an estimated price of about $4,000. Other improvements may include a global positioning system to decrease distance errors and to plan routes. Our system at this time is far from perfect, says Borenstein, but we have a variety of ideas on how to tackle the problems.