Robots could harness AI to master sandwich making

Food manufacturing is the largest manufacturing industry in Europe. However, the industry faces serious problems with labour shortages, in large part due to a shortage of workers willing to take repetitive low-skilled roles. Some repetitive, tightly-constrained jobs can be automated, although machines remain limited in their abilities.

For instance, while Millitec Food Systems supplies robots and other machinery for food production automation, such as assembling and packing sandwiches, they are unable to automate tasks that require machines to differentiate between items, like selecting salad to complete a sandwich.

Millitec paired up with Loughborough University researchers to inject some culinary intelligence into their robots.

Dr Baihua Li and her researchers are developing machine vision and embedded AI technology for the company’s 'Delta' robots, with the aim of allowing them to recognise a range of ingredients. This could allow them to carry out advanced tasks, like picking up the right ingredients in correct quantities and in the correct order.

“This project will deal with various challenges in real-world factory conditions and meet the commercial requirements in accuracy, safety, and speed,” Li said. “The developed robust vision algorithms and camera-based sensing system will also reduce the system manufacturing cost.”

“We will transfer the latest advances in our research in machine vision, deep learning, and robotics to drive innovation for high-care food manufacturing.”

They will use deep learning to develop algorithms to discriminate between ingredients, based on images captured with low-cost cameras. Deep learning is a subtype of machine learning often used in machine vision, in which networks detect patterns from large, unlabelled raw datasets, extracting higher-level meanings from the data through progressive ‘layers’ of processing. In this case, the researchers hope that the robots will recognise various breads and other ingredients, as well as their location and orientation.

It is hoped that the Delta robots will also be able to spot and patch errors on the production line, such as items being misplaced or missing. Eventually, the robots could make and package sandwiches from start to finish without human intervention, allowing human workers to focus on “higher-value” activities.

“We are thrilled to be developing the latest range of food production robotics in conjunction with Loughborough University,” said Richard Ledger, managing director of Millitec. “Throughout our 15 years in the industry, we have seen the challenges presented by skills shortages, and a clear need for a technological solution to support food manufacturers. This development project will be a major leap forward in food production automation.”