We’re pleased to announce that Bytronic has won a new contract to supply our tote inspection system to one of the UK’s largest meal prep brands.
The deal, to supply and install automated deep learning vision systems for an initial two production lines, was signed in November after successful on-site trials. It has the potential to significantly reduce the customer’s energy and water usage when it’s operational next year.
By combining Cognex D900 smart cameras with ViDi deep learning software, the ToteInspect solution automatically detects and diverts dirty totes for washing, removing the need for manual checks and saving unnecessary water use, while learning from experience over time.
“This ToteInspect project lets us help this well-known meal prep company to improve its productivity through AI. It was a real challenge, but that’s what we like. It required not just the latest vision systems and precise lighting, but the right knowledge and expertise in vision from our integration team to make it work. Applying deep learning to subjective inspections like this was, until recently, considered impossible, but new advances in software – particularly learned experience – has opened up whole new possibilities. This latest deal is a great example of this.”John Dunlop, CTO of Bytronic Vision Automation
As products move through the customer’s distribution centre, it’s easy for ingredients to be missed or spilled. The customer needs to ensure totes are clean and empty. Previously, they would have to wash and inspect every tote – a big cost in labour and water.
Bytronic’s vision system uses deep learning technology to automatically recognise and sort clean and contaminated totes on the conveyor line, so that only those totes that need cleaning are washed. This reduces the amount of washing needed and delivers significant savings in energy and water usage.
“This takes subjective inspections beyond the capability of the human eye”
Using Cognex DataMan 260 cameras (above), the system also reads the identification (ID) barcode on each tote to identify its location within the system. The system locks the tote’s ID to its contents inspection image to ensure totes are correctly identified and diverted.
While traditional systems began the automation process, deep learning is enhancing the accuracy and reliability of inspections within face-paced environments such as manufacturing and logistics. This takes automated inspections beyond the abilities of the human eye.
“While traditional systems began the automation process, deep learning is enhancing the accuracy and reliability of inspections within face-paced environments such as manufacturing and logistics. This takes subjective inspections beyond the abilities of the human eye. As well as automated contents inspection, we’re working with other clients on tote type identification, applying the same deep learning methods to distinguish between different physical characteristics such as tote size or colour.”John Dunlop, CTO of Bytronic Vision Automation