Quality Control

The Coolearth Vision System (CEVS) is a solution for increasing the quality and reducing the labor required for production. It replaces what has historically been a labor intensive, time consuming process which only covered a small portion of the product. The vision system does 100% inspection of all articles produced at the speed of the manufacturing line, with a highly repeatable rejection of inferior quality articles. Possible inspection examples include: 1) checking the size and the shape of products (eg: ensuring beef patties are round instead of oval); 2) verifying the color and hue of products (eg: rejection of sub-par vegetables); 3) checking for correct packaging (eg: rejecting candies that lack a wrapper.) Also new forms of QC checks can be realized with thermal (FLIR) cameras that judge either the cooking or the freezing of a product to make sure it has been completely and evenly processed.

Secondarily, when used after a label applying station, the CEVS can check the quality of the label applied. Some of the checks include checking for fading caused by low ink levels, or lines caused by a worn thermal head or clogged ink nozzles. This allows more complex labels to be printed and applied with greater confidence that they are printed and applied correctly. Most of the current print and apply systems rely on a barcode reader to check if the label is applied correctly, with a barcode reader only able to tell if the barcode on the label is readable by a barcode reader. Also when the system detects a decrease in the quality of the printed labels (but still acceptable), it can alert personnel to perform preventative maintenance on the labeling system during the next scheduled down time to reduce unplanned downtime during production.

Automated Grading

The Coolearth Vision System can be used to sort and grade products. Most grading processes that rely on visual or thermal inspection of the product can be accommodated. For example grouping by like color, size or any combination thereof is possible at rates and levels of consistency not practical by humans.

Identification of multiplexed products

In a large production plant many of the items produced are combined onto one final line for palletizing, and for other tasks that are consistent across the product line manufactured in the plant. This in turn requires that each product be pre-labeled with a barcode, identified by human labor, or that very careful timing and synchronization be employed across all of the product lines, so the multiplexed line can tell what product is being processed in each instance. The CEVS eliminates this complexity via optical recognition of each item (distinguishing between them using wording on the container, or their respective colors, sizes, shapes, etc).

Integration with other Coolearth shop floor automation products

Multiplexed product line with Whistle Print and Apply station

By combining the Coolearth Vision Product identification system with the weight and labeling system and label quality system a robust and fully automated system can be realized. Costs can be reduced by eliminating the need for products to be pre-labeled with a barcode before they are multiplexed.

Reduced QA labor costs

The system eliminates the need for QA inspectors. Furthermore it eliminates the need for stand by labor to assist in processing product at times of surge production.

Reduced customer returns

By inspecting 100% of product at each station, customer returns and claims can typically be reduced.

Greater consistency in grading process

For products that require grading having it done with machine vision will result in more consistent, repeatable results. As new grading standards are introduced they can immediately and consistently be applied across all manufacturing lines. The realized improvement in grading allows for a greater understanding of the value and quantity of inventory on hand. This in turn provides for better planing by MRP and APS engines.

Higher quality product

Cherry pick the very best product for super premium offerings. Create a higher margin product out of what you are already processing through the plant.

Hardware

The Coolearth Vision product uses robust, state of the art hardware and software design specifically for challenging shop floor environments. The heart of the standard system is DIN rail mountable fan-less controller that supports up to 100 frames per second of processing and more then 1600 MIPS of calculation. The controller supports 3 Firewire industrial cameras, plus 29 lines of general digital I/O, 5 trigger inputs, VGA output and Gigabit Ethernet connectivity. In addition to processing video the controller can also be used to directly control steppers, servo's, relays and other devices that use TTL or 24v logic, thus it is possible to deploy a full inspection and rejection system with a minimal number of parts and complexity. For the most demanding of tasks, High-Speed industrial panel-mount computers using the latest in Intel and AMD processors, NEMA 4x touch-screen displays and 4 axis full motion control are possible.

Vision Controller Example Inspection and Rejection station A wide selection of Fire-wire, GigE and thermal cameras can be used depending on the application and requirements. The system also includes high intensity, long life (70k+ hours [8 year] run-time) LED lighting to provide for consistent, high quality image capture with minimal maintenance.

Software

Software for the Coolearth Vision system starts with the VxWorks real-time operating system, the same operating system used in elements of the Boeing 787 and NASA’s Mars Landers. On top of this is Coolearth developed software written in National Instruments real-time LabView environment. A blend of custom developed algorithms and industry standard vision analysis lets us deploy high speed, very flexible, inspection and identification systems with short implementation times. The system is easy to learn and use with a GUI that can be custom tailored to the specific application and requirements. Shape, volume, color, OCR, and other types of analysis are all built in.

Example operator status display showing score of objects. The unwrapped candy does not meet all requirements to be scored, and will be rejected, where as the wrapped candy has a score of 91% and will be accepted

Training the system

All visions systems require some degree of training and calibration to the environment and task assigned.

Product Identification

This tasks complexity can vary depending on the state of the products to be identified. If the cases differ in printing and/or labeling the system can be easily trained by placing the product to be identified in front of the system and telling it what it’s looking at. The system is capable of differentiating between several dozen different products at the same time (hardware dependent).