Vision system reliably reads codes on milk cartons

Traceability has steadily evolved from 'nice to have' to 'must have' in the world of food preparation and packaging. Nowhere is this more evident than in companies that supply the major supermarket chains in the UK.

In order to cater for traceability demands – without impacting on production volumes and line speeds – many companies are investing in state-of-the-art sensors and vision systems to help them keep track of not only what is in production, but also where it is going and where it came from. Ultimately, an incorrectly handled package may seem fairly innocuous – but not to the end customer striving to address the needs of an increasingly demanding public.

In one application, at Arla Foods, one of the UK's largest suppliers of fresh milk and cream, Sick (UK) Ltd answered a need when it came to identifying milk cartons destined for a major supermarket. Arla Foods had trialled several other systems, but it was not until Sicks' s IVC2D vision system was put into action that Arla Foods realised that it had found exactly what it was looking for.

The letters 'F' and 'D' may well be next to each other on a keyboard but, in a fast-moving factory environment, they might as well be worlds apart. This, in a manner of speaking, was the issue faced by Arla Foods.

Hopper identification

In its milk cartoning line, two hoppers are used to feed the filling station before the cartons are sealed and ultimately put on pallets for delivery to the customer. To differentiate which hopper filled which carton, Arla Foods prints either an 'F' or a 'D' on the top surface of the carton next to the date code; this is to ensure that each pallet contains one or the other, not a mixture (as demanded by the supermarket chain), in order to maintain its own traceability efforts.

The issue that Arla faced was that both filling stations used the same feed to the packaging centre, so a means of differentiating the cartons was needed that would not only provide accurate differentiation, but also not impact on the speed and throughput of the filling line.

To the naked eye, the orientation of the cartons on the line is uniform, but even a few millimetres or a few degrees out of alignment and the location and orientation of the letters becomes almost random. Combine this with variable printing quality, and it was clear that the IVC2D really had its work cut out. But, after successful trials, it is now an integral part of the production line and is helping Arla Foods pass on the promise of traceability to its subsequent customers.

Location, identification, comparison

David Hannaby of Sick explains: "The IVC2D identifies the position of the last character – the 'F' or the 'D' – and then reads the whole character. It then compares the captured image with the previously taught images stored within the system and matches it to the one that best fits. It then makes a decision on which character it is and sends the appropriate signal to the process controller. It sounds simple but, as you can imagine, the system holds a large number of variants to cater for the alignment and print quality issues; matching the character to one of these, making a decision and passing this information to the process controller has to happen within the blink of an eye."

Steve Brace at Arla Foods adds: "We had trialled a number of other systems but they all had issues related to the product we were scanning. Each carton is the same colour and shape and they often have the same bar code, so what we needed was something that could spot the very minor difference, which, in this case, was actually a major identification criteria.

"It was also important that our throughput was not affected; it is all well and good having an accurate system but, if it adversely impacts your output, you are not really making any headway. In all honesty, there was quite a bit of scepticism, mainly due to the failure of the other solutions, but I am happy to say that some of the earlier sceptics are now the IVC2D's biggest fans.

"Another real bonus is that the software is very easy to use. We have two guys trained up on it and they are capable of making any modifications we need as the process line evolves. We have also had excellent support from Sick; we know that if any issues arise a simple phone call is all that is needed to reach a solution."

Intelligent and flexible automation

The IVC-2D is a high-performance smart camera that enables users to develop intelligent and flexible automation vision systems. Easy to integrate with external devices such as light and trigger sources, it satisfies a vast range of application demands. This ease of integration is largely down to its plug-and-play I/O – compatible with the entirety of Sick's impressive range of sensors and camera accessories. For more complex applications, the camera can also send data, such as object co-ordinates or imagery, over RS485 or Ethernet networks.

Key features include: a 640x480 or 1024x768 b/w sensor, which can operate at 30fps; 10/100MB Fast Ethernet; two program control inputs; three program control outputs; and M12 industrial connectors. There is also the option for a stainless steel enclosure.

The user-friendly IVC Studio programming software gives users quick and easy access to a huge set of powerful image-processing tools. Once configured, the camera can then work in standalone mode, without the need for a PC. Robustness is par for the course, as it is the case with all Sick products, and a protection class as high as IP65 is possible when using the optional lens cap.

Traceability is set to become an even bigger issue across a wider range of industries and it is thanks to companies like Sick, with its broad portfolio of advanced vision systems, that manufacturers can rest that little bit easier.