Article appeared in issue 3, 2021

A research and development partnership between Dawn Farm Foods, Crème Software and University College Dublin’s digital development hub, CeADAR, has secured €1.87m from the Disruptive Technologies Innovation Fund. The funding will be invested in development software used to predict human sensory quality attributes of food products for taste, health benefits and food waste reduction.

The project will use sensor technology to create a digital connection between the consumer, the food scientist and the factory floor to enable real-time product development in a cost-efficient and sustainable manner. This platform aims to reduce the new product development cycle by at least half, transforming how the food industry can accelerate development in a way that reduces food waste and energy and water usage.

 

Conor Cahill, head of science and innovation at Dawn Farm Foods, explains the applications and outcomes of this new software.

It’s a smart, sensory-based, new product development (NPD) platform tool.

Consumers are looking for healthier and more sustainable food products using natural ingredients with less additives or processing aids. Achieving this outcome can be difficult and technically challenging in a manufacturing environment. All food ingredients are naturally variable, both physically and in chemical composition. In any prepared food manufacturing setting, it’s the transformation of these ingredients during processing that is crucial to determining the finished quality of that product.

The software will connect the sensory science of taste, flavour, colour, aroma, and texture through using quantitative rapid sensory analysis with data collected from in-line sensors in our manufacturing environments. This connection between the sensory science and the factory floor will enable our team of food scientists to develop new, healthier, more sustainable products, better tasting products.

To date, companies use data to drive production efficiencies and performance. This application will seek to create a connection to the factory floor using affordable and reliable sensors that will enable a real-time digital twin factory in a cost-efficient and sustainable format.

The project aim is to use recent advances in digital technology to connect our large banks of sensory data from a slice of pepperoni or a plant-based mince, sensory data that is linked to consumer preferences. This new approach will be transformational to the food industry. While the idea of the internet of things and smart machine learning has been around for a long time, we are now taking advantage of this technology by connecting the factory to our sensory platform in real time.

Our team of food scientists will use this platform to develop product formulations and production processes based on the sensory feedback we receive from our customer: This smart machine learning tool will look for insights between the sensory data collected to understand what the optimum settings required are to deliver a set of new recipes that will result in a new low fat pepperoni or a new plant-based mince.

If we successfully connect the sensory data to in-line sensory analysis, the artificial intelligence tools within the software platform will be able to tell us the optimum and precise cooked programme that will deliver a particular texture, colour or even flavour to strict set of sensory parameters while maintaining the highest levels of food safety and confidence in our process. This approach will reduce the time it takes to deliver new product innovations by up to 50% while reducing energy use, water use and food waste.

This will mean that intelligent devices, both in our sensory R&D lab and in our manufacturing environment, can share relevant insights with each other from the start to the end of the process. Ultimately, it allows you to test multiple scenarios for new products and new recipes, enabling us to create products that historically could have taken years to develop. Now, we will be able to do it in a matter of months using this machine learning tool.

This self-learning design tool will offer new and innovative solutions based on production data, sensory data linked to the latest insight-led R&D, and innovation based on the latest consumer trends. This software tool will do the analytics for you, enabling real-time idea generation and recipe selection. The software will give our experienced team of food scientists deep insights and guidance on how to formulate recipes, run virtual factory simulations that consider the needs of all the stakeholders, from raw material selection to what are the likely sensory outcomes for that product formulation when it’s cooked on a pizza, in a sandwich or as part of a meal at home.

If you want to develop a new beef steak strip and your consumer preference is for a particular tenderness and grilled appearance, our team of culinary and food scientist will select the required sensory attributes based on the feedback from this software tool to deliver a more reliable outcome when the product is taken to the factory for trials.

Dawn Farms has a 30-strong team of food scientists and researchers, and culinary expertise with a strong focus on functional and sustainable food innovation. This sensory-based platform is very exciting for our teams: by lowering product development costs, our teams will be able to run multiple new product scenarios to see how each might perform or behave in a factory environment. This platform will create a closed loop of continuous reinforcement learnings that will predict online production outcomes with a high level of accuracy. Traditionally, that could take months or even years of research.

Machine learning is opening the world: computer models deliver intelligence around how your facility works and how it behaves. We believe we are the first in the food industry to use recent advances in digital technology to connect and predict the quality of any finished product. This approach will transform and accelerate new product development using new levels of process confidence, food safety and design capability.

This disruptive technology research project will start this summer – setting up the hardware, installing sensors – effectively creating a pilot digital environment, connecting our sensory science database with real-time sensors suitable for a production environment. It will take time to build, test and validate a scalable industrial software platform that will predict human sensory quality attributes. There will be a large amount of data and modelling required to build and teach this continuous reinforcement learning tool the required intelligence that can deliver valuable insights that will be validated by traditional NPD and sensory science. That is going to take time and we look forward to working with our partners on this journey.

The intellectual property likely to emerge from this programme is a patentable knowledge around using process sensors to predict human sensory outcomes. If successful, this disruptive innovation will ensure any food business will be in a position to develop similar models based on their own data sets and conditions.

We are delighted to be partnering with Crème Software and CeADAR on this pioneering work. The food sector is operating in a dynamic market and the smart combination of culinary science and technology is the optimum way of bringing safe, great tasting and sustainable food to our customers.

 

The Disruptive Technologies Innovation Fund was established by the Irish government in association with Enterprise Ireland under the National Development Plan in 2018 to invest in research and innovation in life sciences, medical devices, ICT, AI, manufacturing, and environmental sustainability. Its purpose is to drive collaboration between Ireland’s world-class research base and industry for the development and adoption of new technologies.