"Harry Barrow (1943– ), a key person involved in the work, later gave this account of FREDDY’s operation:

 

Acquiring an image from the TV camera took quite a few seconds and processing took even longer, and in a single run FREDDY took between 100 and 150 pictures! It took a picture every time it picked up an object to check it has successfully lifted it and not dropped it, and it took a picture every time it put an object down to verify the space was empty. It also scanned the entire world (which required multiple pictures) several times to make a map, and it looked at each object from two different cameras to do some stereo-style estimation of position and size. In fact, the system made the most intensive use of image data of any robot system in the world. The Stanford and MIT systems only took a very small number of pictures to perform their tasks, and relied heavily on dead reckoning and things not going wrong. We, on the other hand, assumed that things were likely to go wrong (objects dropped, rolling, etc.) and made our system highly robust. I really believe that in many ways it was probably the most advanced hand–eye system in existence at the time."

​

From The Quest for Artificial Intelligence, a History of Ideas and Achievements

by Nils J. Nilsson, Stanford University

page 196-197