Page 252 - Introduction to AI Robotics

235
6.7 Range from Vision

Figure 6.22 Uranus undergoing calibration. (Photograph courtesy of Hans
Moravec.)

The major drawback with extracting range information from vision is that
the algorithms tend to be computationally complex. Stereo matching algo-
2
2
rithms are typically on the order of O(n m ). This means that to process an
image of size 640 by 480 takes on the order of 9 10 10 instructions, while
to simply segment a color is on the order of O(nm ) or 3 10 5 instructions.
Even with advances in microprocessors, a stereo range map can take minutes
to compute.

6.7.2 Light stripers
Light striping, light stripers or structured light detectors work by projecting a
colored line (or stripe), grid, or pattern of dots on the environment. Then a
regular vision camera observes how the pattern is distorted in the image. For
example, in Fig. 6.26a, the striper projects four lines. The lines should occur
at specific, evenly spaced rows in the camera image if the surface is flat. If
the surface is not flat, as shown in Fig. 6.26b, the lines will have breaks or