Many manufacturing plants use vision systems to monitor and control production processes. These systems most often are supplied with machines and robots purchased by the plant, although they are sometimes retrofitted onto existing manufacturing lines by OEMs or plant personnel.
Vision systems tend to encompass a wide range of applications, leading to many different types of systems in a typical plant. Even if all the vision systems in a plant are provided by the same vendor, there usually will be a wide range of model types and custom configurations. This wide range of systems is necessary to optimize performance, as it's rare for any two vision applications to be identical.
SEE ALSO: Bring Closure to Product Defects
Most vision systems are comprised of a camera on the front end, a controller in the middle, and a PC-based operator interface platform on the back end. Many vision system suppliers can provide PC-based software along with their vision systems to allow visualization, storage and analysis of data acquired from one or more cameras.
In some cases, the camera is combined with the controller to reduce the size and cost of the vision system. Separating the camera from the controller is a better choice in other applications, particularly those in which the camera is subject to in-service wear and tear.
In most applications, the vision system is designed to determine a specific condition. Examples are search (detection of presence of an item such as a label), edge pitch (the number of edges in a region), and edge position (detection of edges and their positions).
Most suppliers offer a very similar product, with differences in performance among vision systems determined by lensing, lighting and other application-specific factors. Some of the more important basic terms in a vision system application are field of view (complete area of inspection), region of interest (specific area of interest), inspection area (precisely where a particular feature is being detected or inspected), and resolution (defines the theoretical smallest detectable object per a given field of view and as a function of a certain resolution camera).
"Resolution is often misunderstood as a value determined solely by the camera, when in fact it is a function of actual pixel count of the camera and the field of view," says Rob Fell, strategic account manager at Omron. "For example, resolution can be increased without investing in a higher pixel camera by decreasing the field of view. It's often possible to focus only on a key part of the object, enabling a small field of view, and a consequent high resolution."
Once the right camera/controller hardware is purchased and configured, the next step is to decide how to visualize, store and analyze the data. A low-cost solution is to use the PC-based software supplied by the vision system supplier. For a factory with very few vision systems with a limited interface to the main automation system, this could be a feasible alternative.
But for manufacturing facilities with many vision systems, and even for those with just a few vision systems that must be tightly integrated with the main automation system, a better choice is often a SCADA system. "Connecting all the vision systems to a single SCADA system provides consistency in interface, streamlined recipe control and centralization of information," notes Lourenco Teodoro, vice president of engineering at InduSoft.
The selected SCADA system must be able to interface with multiple vision system controller platforms via standard interfaces such as various versions of Ethernet, and also via more esoteric protocols specific to vision system hardware.
"The simplified user experience of one common SCADA platform makes it easier to store and analyze vision system data, and to use this data to perform custom adjustments that can improve visualization of operations. These adjustments can include zoom in and de-cluttering of the image to optimize data presentation, actions made easier with advanced features such as multi-touch technology," Teodoro adds.