Packaging has many challenges when it comes to automated production – and some processes are ill-suited. Flat-pack furniture manufacture, for example, is not usually associated with automation, since it requires the correct selection of all the component parts and accessories. Typical quality problems include missing bags of screws, missing panels, wrong types of panels, missing instruction sheets and so on.
Acre is one of the leading manufacturers of flat-pack furniture and is leading the way in highly automated quality control. Industrial Vision Systems has developed an automated inspection system for Acre (pictured in action above) that will check the components of the flat pack prior to sealing and final delivery of the product. Pre-requisites for the system included a rapid and configurable user interface, coupled with the ability to show real-time feedback to the operator for problem parts, and trends emerging from the manufacturing process.
The main problem was the large variety of parts it manufactured, and the infinite number of styles and types for various customers. Therefore, it was important that the system had the ability to store and retrieve pre-set inspection sequences.
The packs to be inspected are driven along a continually moving conveyor. Operators stand to the side of the conveyor and carry out operations as the pack moves past their positions. Towards the end of the line is the vision system, as shown in Fig 1.
As the parts enter the machine, a sensor is ready to detect the position of the pack. Once the workpiece is centred in the machine, the sensor is made and the PLC triggers the vision system for the inspection to take place in real time. The vision system is based on NeuroCheck software in conjunction with NeuroCheck FWXC13c FireWire cameras, which offer huge benefits to Acre for this sort of inspection. With the NeuroCheck FireWire technology individual camera set-ups can be allocated and changed in real time depending on the pack requiring inspection and the particular light requirements for that pack.
Two high-resolution FireWire cameras are used, with colour sensors offering the ability in NeuroCheck to perform a number of differing image processing techniques, including colour matching, template matching and neural network classification in key areas of the furniture.
The pack does not have to arrive parallel to the track as, through NeuroCheck, automatic position adjustment is performed. The inspection takes place in real time, while the pack continues to move along the conveyor. If a failure occurs, the line is stopped automatically via the PLC, the nature of the error is shown through the NeuroCheck software screen and, finally, an audible siren alerts an operator to correct the pack.
An important part of the whole system is the reliable and exact nature of the lighting and optics used – giving repeatable and dependable results. For the system, two high frequency machine vision fluorescent lights are used in conjunction with polarising filters. The optics for the system were 50mm manual iris, fixed-focus lockable lenses.
NeuroCheck has the ability to sub-divide the inspection criteria into discrete steps (called ‘individual checks’) which make up a ‘check routine’. It was important to methodically build up the database of good and bad parts by constant testing through the development architecture that NeuroCheck offers, i.e. the ability to test results on the same system that will later be used in production.
The unique ability to offer flexible inspection criteria, in tandem with the ability to save data to Excel in real time, gives the system ultimate power for detailed analysis of the failures. The software will collate statistical information for the individual errors, as well as saving individual images of the failed parts to give a visual database of any errors occurring.
The total solution gives unparalleled results in the quality inspection of flat pack furniture, offering a never before achieved level of defect detection. The system is characterised by simple, clear operation and a high degree of reliability. Changes to parameters are password-protected on the shop floor, and the system is networked via an Ethernet connection to the machine. Therefore, any new parts added to the system can be set-up via engineers from the comfort of their desks.
Battling with bottles
Ottakringer Brauereu AG in Vienna is using visual inspection to check that beer crates and boxes of canned beverages contain the correct amount of product. The second-largest Austrian brewery was able to make significant production and quality improvements while reducing costs by using the ‘Checker 101’ high-speed sensor from Cognex. Image-based sensors are increasingly taking over the tasks previously performed by standard sensor technology. In this brewery, intelligent sensor technology benefited production lines and sales and also proved to be an important factor for the company’s activities in the competitive contract-bottling market.
The main priority of the brewery’s maintenance department is the inspection of beer crates and boxes of cans to ensure they contain the correct amount. Ottakringer contacted the image processing department of Cognex partner Schmachtl GmbH. The brewery’s production automation needs and priorities were established by the Schmachtl team. These tasks are performed by Checker 101. In order to allow the brewery to develop an in-house solution, Schmachtl provided a test unit and specialist advice giving the maintenance team the flexibility to set up the checking station themselves. The ease of installation and operation, as well as the simple programming of the system using a laptop computer, proved significant advantages. The brewery was able to start automatically inspecting the boxes of beverage cans in January 2006. 
The Cognex Checker 101 vision sensor in use at Ottakringer Brauereu in Vienna.
The procedure involves passing each box along a running conveyor belt equipped with the Checker sensor, to determine whether it contains 24 correctly inserted cans before the box is shrink-wrapped. The results of the check are transferred directly to the production control system. Production flow is improved and batch production statistics are easily obtained.
Their positive experiences with the checking system for boxes of cans encouraged the maintenance department to introduce the solution to further applications. The next step was to introduce a completeness check on the beer crates. This inspection method needed to be replaced with a more modern, more flexible and more reliable checking technology. In the past, a complicated system technology using many individual sensors was used. The in-house maintenance department decided to use Checker in February 2006 and the job was complete by March. The simple, space-saving design proved to be of great benefit, meaning that no fundamental changes to the conveyor line were required. Checker inspects the beer crates on the running conveyor belt to check they have the correct amount. Checker can also differentiate whether the bottles have light or dark tops, thus performing an extra quality inspection on the bottling line. The intelligent sensor automatically adjusts to the respective product, determining whether an 18-, 20- or 24-bottle crate is on the line. The operating staff do not have to modify the checking station when changing the job type. The checking station now operates on
a two- to three-shift basis and reliably inspects beer crates at a rate of approximately one per second. Integration of Checker into the control technology of the production system was easy due to the standardised interface. Checker also causes the production belt to stop automatically if a fault is registered.
An intelligent solution that means quality improvements are clearly noticeable, right down to customer level, ultimately contribute to an improved company image and the brewery is already checking the possibility of additional applications.