Improving inspection with machine vision

04 April 2022

Food Processing asked how machine vision can improve inspection processes in food production environments and what engineers need to consider when specifying machine vision on their lines.

Machine vision is a powerful and versatile tool whose potential to improve food production efficiency has grown tremendously in the past few years. Advances in both hardware and software have made vision more accessible, easy to set up and operate for food processors. Vision has developed not only to give inspection processes the ‘eyes’ to measure accurately, to detect subtle ranges of colour and shapes, but also the intelligence to classify products with infinite differences, as well as to read and verify labels. Inspection has also been integrated into pick-and-place robot guidance systems.

“While greater accessibility has led to increased adoption, there are still plenty of food processors who continue to rely on humans to check their products at the end of the line, even though human inspections are less reliable and repeatable,” said Neil Sandhu, chair of the UK Industrial Vision Association and product manager for imaging, measurement and ranging at SICK UK.

The good news

“The good news is that the capabilities of enhanced 2D, 3D and even artificial intelligence (AI) inspections have become more sophisticated, but also more affordable – powered by hardware that packs greater performance into smaller devices to fit into tighter machine spaces. However, it is in the software that the greatest advances have come – with more systems now offering the benefits of rapid configuration for common food production tasks, with some available ‘out of the box’ as ready-to-use packages.”

Sandhu went on to point out that vision systems routinely inspect both the product, and elements of the production process itself, either to control a process – for example sorting different coloured yogurt containers – or identifying potential production faults, such as looking for residue in chocolate moulds to avoid overfilling.

When specifying machine vision on a production line Sandhu advises that the first step should be to drill down and understand exactly what it problem needs to be solved. “Start by focusing on the most important aspect in the production process and be clear about what is a pass or fail. Vision systems can then evolve and grow,” he said. “Make the environment robust and repeatable, so the same result can be expected every time. It’s all too easy to end up with a ‘white elephant’ system that gets turned off because it doesn’t really do what you want it to do.”

Sandhu goes on to advise that inspections should be positioned as early as possible in the production process, otherwise value is added unnecessarily, creating extra waste. He said: “Traditionally, human inspections have been carried out at the end of a line, but why add icing to a cake that is too high, or chocolate to a mis-shaped biscuit? Then inspect it when it goes in the packaging?”

Inadequate inspection processes can reduce yields by creating unnecessary waste and longer production stoppages while problems are identified and corrected. Inspection processes also have a vital role to play in optimising Overall Equipment Effectiveness (OEE) – a measure of production efficiency that is increasingly being adopted by manufacturers. “The return on investment of a vision system must be evaluated in terms of increased yield,” concludes Sandhu.