VISUAL ROBOTS.

It is shown that it is possible to automate the processes of visual observation, recognition, and noncontact measurement, and that this automation can be effected at speeds (1000 frames per second) and accuracies (0. 1 micron) far beyond the capabilities of a human. This automation uses image scanning which assigns short simple quantitative signatures to image shapes. The signature for each image shape is unique and unambiguously defines the specified shape for the system. The technology involved is outlined. The existing machines which measure and memorize image information on command and then recognize and adapt to image shifts automatically at speeds much faster than a human eye can follow are described.