Page 112 - Concise Encyclopedia of Robotics

Error Accumulation
observing the island and scrutinizing the shape and angular size of the
island’s image.
As the drone flies along, the island seems to move underneath it. A
camera is fixed on the island. The controller sees an image that constantly
changes shape and angular size.The controller is programmed with the true
size, shape, orientation, and geographic location of the island. The con-
troller compares the shape/size of the image it sees, from the vantage
point of the aircraft, with the actual shape/size of the island, that it
“knows” from the map data. From this alone, it can precisely determine
the drone’s:
•Altitude
• Speed of travel relative to the surface
• Direction of travel relative to the surface
• Geographic latitude
• Geographic longitude
The key is that there exists a one-to-one correspondence between all points
within sight of the island, and the angular size and shape of the island’s
image. The correspondence is far too complex for a human being to
memorize exactly; but for a computer, matching the image it sees with a
particular point in space is easy.
Epipolar navigation can, in theory, work on any scale, and at any
speed—even relativistic speeds at which time dilation occurs.It is a method
by which robots can find their way without triangulation,direction finding,
beacons, sonar, or radar. It is, however, necessary that the robot have a
detailed, precise, and accurate computer map of its environment.
See also COMPUTER MAP, LOG POLAR NAVIGATION, and VISION SYSTEM.
ERROR ACCUMULATION
When measurements are made in succession, the maximum possible
error adds up. This is called error accumulation.
Analog error accumulation can be illustrated by a measurement example.
Suppose you want to measure a long piece of string (about 100 m, say),
using a meter stick marked off in millimeters. You must place the stick
along the string over and over again, about 100 times. If your error is up
to ±2 mm with each measurement, then after 100 repetitions, the possible
error is up to ±200 mm.
Digital error accumulation occurs as bits are misread in a communica-
tions circuit, incorrectly written on disk, or incorrectly stored in memory.
A machine might see a logic low when it should see high, or vice versa.
Suppose that, for a particular digital file, an average of three errors are
introduced each time the file is transferred from one node to another in

