Biomedical Engineering Reference
In-Depth Information
FIGURE 7-16
Guide cane
prototype
(a) Schematic of the
operational
principles.
(b) Photograph of
the prototype in
operation. [Adapted
from (Borenstein and
Ulrich 1997) with
permission.]
forward. This obstacle map is then used by the onboard computer to calculate the best
direction to travel, even through a cluttered environment. If an obstacle is detected in the
path, the steering servo drives the angle of the GuideCane wheels in a direction to avoid
it. The operator feels the changed resistance on the cane even before the direction change
is detected and intuitively follows the direction suggested with almost no hesitation.
The GuideCane solves the problem of stairs in two ways. When a drop-off is reached,
the wheels drop off the edge providing a fail-safe indication of the fall. In the other
direction, the main forward-looking array sees the bottom step as an obstacle, but the
forward-looking up-facing sensor measures a longer range. If the difference between the
two ranges is about 300 mm, the object is treated as a flight of stairs. However, if the two
distances are almost the same, then the object is treated as a wall and is avoided.
GuideCane incorporates two additional functions: (1) a side-looking sonar for wall
following; and (2) a global positioning system (GPS) that can be used for navigation of
preprogrammed routes. The latter are generated automatically during a “lead-through”
run under the guidance of a sighted person.
For indoor navigation, dead reckoning based on compass readings and wheel-sensor
odometry can be used for short distances, but the integrated error grows too large to achieve
the required accuracy for distances farther than a few meters. However, algorithms such as
simultaneous localization and mapping (Leonard and Durrant-Whyte, 1991), a well-known
algorithm used by the robotics fraternity, could be implemented to solve this problem.
