You might not think it, but 70% of factory tasks are still performed manually and 90% of them involve a visual component. The majority of these control and monitoring tasks do not often add value for operators and could be automated most of the time. However, until now machine vision technologies have been complex and costly to implement. Automi, a startup co-founded in 2021 by Galem Kayo, has devised a solution that now opens the way for large-scale automation of visual inspections thanks to artificial intelligence.
A Visual Inspection Automation Platform
“We’ve chosen a ‘no code’ approach,” says Galem Kayo. “We give operators access to artificial intelligence through digital tools that are very easy to use. This means they can train their own robot without having to write a single line of code.” For example, to train a quality control algorithm, the platform uses images of non-compliant parts. The operator circles their defects, as if using a sheet of paper, and adds a descriptive label to distinguish them. The robot gradually learns to recognize all faults automatically. “One of our customers, an aircraft part manufacturer, checks an additional 3 million parts every year, without hiring or training any operators.”
One of our customers, an aircraft part manufacturer, checks an additional 3 million parts every year, without hiring or training any operators.
Automi also helps to automate production line performance levels. A leading European baker uses the platform to assess production pace in real time. The trained algorithm measures the production output and raises the alarm if it detects any anomalies, increases or decreases. It is an effective way to produce the right amount without generating waste.
5G and Edge Computing: The Ideal Combination
Automi processes large volumes of data (mostly high-resolution images) that need to be transferred via broadband with high bandwidth and no latency to allow decisions to be made instantly. 5G comes into its own here, as it makes it easier to process mass data in real time.
Similarly, to train and guide its robots, the startup is interested in edge computing — decentralized cloud technology that can process large volumes of data with low latency, thanks to how close the computing power is to the customer’s installations. “We currently work from embedded computing systems on our customers’ premises, which forces us to manage the configuration and maintenance of these devices. In future, we will process data locally, at the edge of the network — a solution that should save us time and reduce costs.” Edge computing is also a real asset in providing tailored solutions for industrial customers who consider privacy and data protection to be essential prerequisites. These manufacturers prefer to have their data processed as close as possible to its source, on their sites.
Urgent Needs in All Sectors
It’s clear that there is a great need for this. In barely a year, Automi has already reached double digits for the number of pilot projects that it has converted into sustainable devices in the aeronautical, automotive, food processing, textile and pharmaceutical industries, etc. Its accessible solution quickly becomes operational in all companies that need to increase their throughput and improve their performance but struggle to recruit and retain operators.
