Cognex Elevates Machine Vision with AI-driven Edge Technology
Cognex has released its latest in machine vision, the In-Sight 2800 Detector, with artificial intelligence on an edge device, cutting out the need for expensive networking and computing hardware.
When it comes to machine vision in automation, Cognex is well-known around the world for having a wide variety of machine vision products, from barcode readers with liquid lens technology to AI-driven vision processing computers. Recently, Cognex added the In-Sight 2800 Detector, an AI-based vision detection system that uses edge learning, to its offerings.
Cognex’s new In-Sight 2800 Detector uses AI and edge learning technology for machine vision in logistics. Image used courtesy of Cognex
In-Sight 2800 Detector
As online shopping grows in popularity, the logistics industry is turning to automation to reduce wasted time getting products from the shelf to customers' hands. Built with the logistics industry in mind, the In-Sight 2800 Detector is a vision system that can detect conveyor jams, detect items in totes or on sorters, classify packaging types, and identify other process issues.
To reduce the integration time of vision systems, Cognex uses AI to detect objects. This AI algorithm uses a small image set so the amount of time teaching good or bad images is greatly reduced. Unlike other AI-driven vision systems, however, the In-Sight 2800 Detector uses edge learning tools on the device or "at the edge." This means the image is processed directly on the device instead of processing the image on large computers that may not be suitable for industrial use. Because of the edge technology, no external hardware or advanced programming is needed.
The In-Sight 2800 Detector uses the same AI technology as typical AI-based vision systems but does all the processing on the device rather than needing a PC. Image used courtesy of Cognex
AI-driven Vision Systems
A typical AI-vision system requires cameras and a PC. The PC will need to be mounted on the line someplace secure and clean. Power over Ethernet (PoE) or Ethernet cameras are often used with gigabit switches and high-speed cables back to the PC. The camera images are processed in the PC, with results sent to the PLC.
While this configuration does work, the process can be costly and requires time for setup and configuration. The In-Sight 2800 Detector uses the same AI technology as a standard AI-vision system, but does all of these processes on the device. The camera only needs to be connected to power and communications to start snapping pictures.
The In-Sight 2800 Detector can be used in a range of applications, including process issue detection and item detection and sortation. A process issue could be part related or a conveyor jam—either scenario can be difficult to detect with a typical vision system. By teaching the system what consists of a good part and a bad part, however, the In-Sight 2800 Detector can easily sort the good parts from the bad.
The In-Sight 2800 Detector can spot objects regardless of the object's background color or pattern. Image courtesy of Cognex
When it comes to item detection, other vision systems might use a pattern match, but with low contrast between the item to be detected and the background, using AI technology can help reduce the number of false negatives or false positives. With item sortation, classifying objects can sometimes be difficult for a vision system. With the help of AI, however, each classification can be taught, making it easy for the vision system to detect the proper classification.
On the Edge
Being able to quickly deploy an AI-vision system with minimal impact on existing systems sets the In-Sight 2800 Detector with edge learning technology apart from standard vision systems. With its ability to detect and sort devices with high throughputs, Cognex's new offering is expected to see primary use in the logistics industry.