Machine vision exposed
Machine vision has so embedded itself in the manufacturing process that it may be seamless to the untrained eye, with end results being equally undetectable.
Machine vision has so embedded itself in the manufacturing process that it may be seamless to the untrained eye, with end results being equally undetectable. That is certainly true of the vision system detailed in our cover story, which is used to calibrate a drilling robot installing fasteners on the tailfin vertical stabilizers of an F-22 Stealth aircraft. The system described by system-integrator Roger Richardson of Delta Sigma and under test by Lockheed Martin ensures that not one fastener protrudes above the surface of the aircraft skin. Otherwise, the pilot could be exposed to enemy radar.
Many applications for machine-vision technology are not of such an immediate life-and-death nature. However, as machine-vision systems become a critical part of an increasing number of industrial-automation systems, greater rewards from higher-quality goods or faster system throughput become the norm. In this issue, system-integrator David Wyatt of Midwest Integration describes a system that his company designed for Mittal Steel, one of the world’s largest steel producers. The linescan-camera-based system sits in a stainless-steel enclosure below a conveyor and inspects long, intensely hot strips of steel, resulting in increased output quality and quantity as defective strips are identified, reworked, and sold.
Similarly, improving the quality and quantity of fish to consumers is also a task now being performed by machine-vision systems. As contributing editor Charlie Masi reports, automating the design of a fish-sorting station on a commercial fishing boat enables the right species and quality of fish to be fed into a filleting machine in the right orientation and at high speed. Based on a neural-net computer, the system identifies the fish and accepts, recycles, or rejects them, with the result that the ship’s hold is filled faster and consumers are delivered fresher fish.
Smart cameras uncloaked
“Smart camera” is a term often used by manufacturers to denote a deceptively simple machine-vision component that incorporates all of the features of a system in a single camera. In his article on these vision sensors, editor Andy Wilson describes the cameras and sensors, processors, operating systems, and interfaces that are used in their designs. Despite the fragmented nature of these designs, the cameras can indeed be highly useful in many machine-vision system designs.
As part of our ongoing discussion of machine-vision components, author David Berg of Oren Sage Technology describes how to best choose lenses for machine-vision applications. He provides a clear method for evaluating lens parameters and selecting an appropriate lens for an application. To find a more complete discussion of these issues, David Berg’s webcast, first broadcast by Vision Systems Design in August, is now archived on our Web site (www.vision-systems.com). As far as machine vision is concerned, we have no intention of keeping the secrets of successful system design in stealth mode.
W. Conard Holton
Editor in Chief