Page 253 - Introduction to AI Robotics

236
6 Common Sensing Techniques for Reactive Robots

b.

a.

Figure 6.23 Robots and stereo: a.) The Stanford Cart developed in the late 1970’s.
(Photograph courtesy of Hans Moravec.) b.)The Marsokhod rover developed in the
late 1990’s jointly by scientists from McDonnell Douglas, Russia, and NASA Ames
Research Center. (Image courtesy of the National Aeronautics and Space Adminis-
tration.)

discontinuities. A vision algorithm can quickly scan each of the designated
rows to see if the projected line is continuous or not. The location of the
breaks in the line give information about the size of the obstacle. The vision
algorithm can also look for where the dislocated line segments appears, since
the distance in image coordinates is proportional to the depth of the object.
The relative placement of the lines indicates whether the object is above the
NEGATIVE OBSTACLE ground plane (an obstacle) or below (a hole or negative obstacle). The more
lines or finer-grained grid, the more depth information.
Light stripers are less expensive for many reasons. First, since they are pro-
ducing a line or pattern to be measured, expensive time-of-flight detectors
are unnecessary. The detection is done by the camera. Second, producing a
thick line that can be detected by a camera does not require a laser. Instead it
can be done with structured light, an optical method which allows “regular”
light to mimic desirable properties of laser light. Finally, light stripers pro-