PATTERN RECOGNITION -21101

In a robot vision system, one way to identify an object or decode data is by shape. Bar coding is a common example. Optical scanning is another. The machine recognizes combinations of shapes, and deduces their meanings using a microcomputer. In smart robots, the technology of pattern recognition is gaining importance.Researchers sometimes use Bongard problems to refine pattern-recognition systems.

Imagine a personal robot that you keep around the house. It might identify you because of combinations of features, such as your height, hair color, eye color, voice inflections, and voice accent. Perhaps your personal robot can instantly recognize your face, just as your friends do. This technology exists, but it requires considerable processing power and the cost is high. There are simpler means of identifying people.

Suppose your robot is programmed to shake hands with anyone who enters the house. In this way, the robot gets the fingerprints of the person. It has a set of authorized fingerprints in storage. If anyone refuses to shake hands, the robot can actuate a silent alarm to summon police robots. The same thing might happen if the robot does recognize the print of the person shaking its hand. This is a hypothetical and rather Orwellian scenario; many people would prefer not to enter a house so equipped. However, that fact in itself could arguably serve as a security enhancement.