Industry first labelling assurance system that uses RFID tracking technology to accept and reject products by inspecting the labelling on every package to ensure accuracy
Using our advanced innovation solution architecture to effectively implement the solution in a mission-critical, 24/7 environment without impacting production
Designed and implemented an innovative, Australian-first machine learning project within a four month timeframe
Manufacturing, Food processing
The Client is a large Australian agricultural processor supplying premium produce to domestic and export markets.
As one of the largest processors of its type in Australia, the Client prepares, processes, and packages its products and distributes it via wholesale, retail, and eCommerce platforms.
In the midst of the COVID-19 pandemic, in an already challenging commodity market, the government in a major export market of our Client’s raised concerns around a small number of product labelling non-conformances.
The non-conformances were not due to product safety or quality, but because some of the labelling stated a production date on the exterior packaging that did not match the internal product labelling production date of one day earlier. This was due to the 24/7 nature of the product, as the production date would change as products progressed down the factory line.
Non-conformances were also found related to the labelling versus the actual contents – for example, the label may have stated that the contained four of Product A when it actually contained three of Product A and one of Product B.
The need to look ‘in the box’ posed a unique problem: how do you see what's on the label and what the product is inside a box when the box is sealed?
Clevvi was engaged to design a solution that delivered on the Client’s vision to validate their product packing and labelling activities in real-time and on an automated basis that scaled up to tens-of-thousands of boxes per day, ensuring that the contents of the packaging matched the label.
Working closely with the Client’s IT and engineering teams, on-floor/MES system provider, and RFID provider, RAMP RFID, Clevvi designed and implemented a system that applied image recognition and RFID technology to 'read through' the labelling and ensure it was carried out as planned.
RFID works like an electronic bar code. When combined with image recognition and record matching, the technology enables automation of inspection processes at scale, and delivers a level of accuracy which would not be achievable through human inspection processes.
Running the project with Agile methodology, Clevvi created the technical specification and designed the integrations to scan and validate the product. Further, we built the decision support system that matched product data to RFIDs and assessed whether the box was conforming or not.
The architecture that Clevvi designed and implemented uses a microservice architecture and containerisation. This means that a small component can be changed without impacting other parts of the system. As the Client’s factory runs 24/7 and only has four short planned shutdown maintenance windows each year, it was essential that the design was flexible enough to be carried out while live in the production environment.
Clevvi architected, implemented, and deployed the system through the life cycle of a proof of concept (POC) through to a full-featured innovation project and finally, upon the value having been proven, commercialisation to be used permanently by the business at scale.
Clevvi included in our design architecture specialised cameras on the conveyor belts above where the products travel to take photos and create a visual history for every package. This is also utlised in a machine vision context.
The Clevvi team can act as an outsourced innovation partner for technical operations or production environments.
We deliver innovation as a service by working first to analyse the functional requirements, then build the architecture and integrations to existing on floor systems. We can bring machine learning and IOT technologies into mission critical 24/7 environments whilst managing the risk of introducing it through careful architecture and deployment.
Within four months, Clevvi was able to deliver the Client’s vision to validate labelling compliance in real-time and at scale. The technology allows the Client to successfully track whole and part products from producer to end-customer, reducing the margin of error and issues surrounding non-compliance in export markets.
The new system was successfully installed within the planned maintenance shut down window and, owing to the microarchitecture, the Client does not need to stop the production line to make additional changes.
The innovation project proved to be a commercial success and created demand amongst clients overseas. Clevvi received a subsequent engagement to help the client extend and industrialise the solution as part of its commercialisation.
Clevvi’s work was part of a patent pending, Australian-first machine-learning project that has demonstrated the role that technology can play in achieving automated quality inspection and assurance. Through collecting thousands of images throughout the process, the system is now learning to perform accurate Machine Learning (ML) image recognition. It has the potential to perform automated visual inspection of products, including which type of product is being sent out and if it is in line with specific customer requirements.