Bringing real-time Edge AI recognition to a high-volume consumer product

A client wanted its consumer device to recognise different product variants automatically, using their existing markings, without adding cost to each consumable. This would create a foundation for higher-value product features while avoiding the need for additional tags, codes or other features. 42T designed, built and tested an Edge AI proof of concept that processed each reading locally, providing rapid recognition without dependence on the cloud.

Key outcomes

The challenge

A global consumer-goods manufacturer wanted its next-generation smart device to recognise automatically which product variant a user had selected. Recognition needed to use the markings already present on the product, without requiring an additional tag, code or other feature that would increase the cost of every consumable. At the volumes involved, even a small increase in the cost of each consumable would be commercially significant.  

Reliable recognition could allow the device to adjust its operation automatically for different product variants, helping create a simpler, more personalised consumer experience and providing a foundation for future connected services. The recognition hardware needed to sit within the reusable device and meet a demanding module cost target. To become a viable production feature, the system would need to achieve a very low error rate, accommodate variation in product placement and distinguish between a broad and varied product range. Before committing to a production design, the client wanted greater confidence in the proposed sensing approach and a clearer understanding of the development required to progress it.

42T’s solution

42T designed, built and tested an Edge AI sensing solution that used optical measurements rather than camera-based recognition. The team combined low-cost sensing hardware with a lightweight classification model running entirely on-device, enabling products to be identified in real time without relying on cloud processing.

The approach addressed the client’s demanding cost and performance requirements while keeping the recognition technology within the reusable device, avoiding the need to add cost or complexity to every consumable.

The team developed the sensing and classification system into a working demonstrator, refining the sensing arrangement to improve signal quality and measurement consistency. The system was then tested across a broad sample of products to establish recognition performance and identify the main causes of remaining measurement variation.

By separating the effects of the prototype design, product positioning and manufacturing variation, 42T identified where further engineering effort would have the greatest impact. This gave the client an evidence-based understanding of the technology’s potential and limitations, together with a prioritised route towards production-ready performance.

“Commercially constrained problems such as this are very typical of the work we do for our clients. The application of Edge AI often has a part to play to open up solutions to problems that previously did not exist. The team combined low-cost sensing with intelligent processing at the edge to demonstrate reliable, real-time product recognition without adding cost or complexity to the consumable. They also identified the factors limiting performance and gave the client an evidence-based route to production.”

Why 42T?

Meet the team

Stuart Gilby

Stuart Gilby

Director of FMCG

Stuart Gilby is Director of FMCG at 42T, with nearly 15 years of experience in technical consulting, product development, and innovation strategy. He specialises in developing new products, packaging, and production processes for major FMCG brands and ambitious start-ups.

Simon Jelley

Simon Jelley

Principal Consultant – Head of Innovation

Simon is Head of Innovation at 42T and a Chartered Mechanical Engineer with a Master’s in Engineering Science from the University of Oxford. Simon thrives leading innovation programmes that span thermofluidics, mechanism design, sensing systems, and systems-level thinking.

Read more

line clearance tech demonstrator

Integrating Edge AI for faster pharmaceutical line clearance

42T designed and built a working Edge AI demonstrator for pharmaceutical line clearance, combining an…

How Edge AI is transforming cosmetics manufacturing

Cosmetics manufacturing combines complex formulations with high production speeds and exacting expectations around quality. A…

42 Technology and Omnisense collaborate on safer autonomous drone landing system

42 Technology (42T) has played a central role in helping Omnisense, a company specialising in…

Let us help you

We excel in deep innovation and technical breakthroughs, from early-stage exploration to end-to-end development and manufacturing.