Division President, Shane Mellor, Managing Director, Mellor Metals (GBR) set the agenda by asserting “AI is fast becoming a critical tool for recyclers”, joking that AI-driven robots would soon “be complaining about the commodity prices”.
“AI can help us detect hazards like lithium-ion batteries before they cause damage, improve operational efficiency, raise safety standards and support the circular economy strategies,” he said. “Steel recyclers don’t wait to be told what to do – we innovate, adapt, and lead.”
Two guest speakers then set out the range of tasks AI can perform in routine yard tasks and boost business efficiency. Raghav Mecheri, founder and CEO, Visia (USA) explained how his company combines AI and X-ray technology to identify materials – calling it “material intelligence”.
“Feedstock volatility, or the unpredictability in raw materials coming into yards is really, really hampering yard operations,” he said. Highlighting fires caused by batteries, Mr Mecheri said he had seen several but those at recycling yards were “categorically the scariest”.
“Beyond that, we also see feedstock unpredictability and dirt coming into yards. How do we know how much of a load of ferrous metal is truly iron?”
Manual inspection alone, he asserted, was not the best way and described his company’s role in deciding which equipment met the specific needs of the client, including cameras to identify value in fines or to spot hazardous items. X-ray technology is used to “see through” loads to assess contamination or to spot lithium-ion batteries hidden within metal products. It was possible to identify the chemistry of 65 different battery types.
Mr Mecheri claimed that combining the sensors with AI delivered up to 97% accuracy in spotting batteries, meaning some training and safety protocols were still needed. “No-one can eliminate catastrophic events entirely.”
The second speaker, Sean Davidson, founder of Jules AI (CAN), called the fusion of AI, sensors and robotics “the single biggest measurable commercial win”, offering economic benefits, operational excellence and environmental and safety impacts. He said ROI was typically achieved within 12-24 months. Jules AI offers software for the recycled materials industry, with an emphasis on international trade. It covers automation, risk management and market intelligence. Mr Davidson listed key players in the industry making major use of AI or machine learning: AMP Robotics, Zen Robotics, Tomra Sorting and Steinert.
“Danieli have a fully autonomous yard at one of the steel mills [in Canada] which will absolutely blow your mind because it not just classifies the material as it comes in, they have automation that can actually tip and tilt.”
In a discussion which followed the presentations, the panel was asked whether the costs of investing in AI would mean the end for smaller operators. Ben László, Head of Information Technology Asset Disposal, Kuusakoski Recycling (FIN) advised SMEs to consider different types of financing models. “It’s maybe not the most feasible [course] for SMEs to purchase a device but look into leasing models, ambassador models or cooperation models. You really need to get the details correct if you want to win the game.”
“I would argue these new system providers are looking for partners that really have the effort and the interest to put in the time. SMEs have a niche to penetrate.”
Mr Davidson agreed, saying providers priced their solutions in recognition of SMEs. “They have different pricing models – you can lease it; you can pay per scan. This isn’t just something that’s accessible only to the bigger players. It’s already accessible at an SME level.”
Moderator George Adams, CEO, SA Recycling (USA) was concerned about the potential danger in the workplace of radiation from X-ray sensors. He was assured by Mr Mecheri that “radiation compliance is a consideration we have to take into account for everything we deploy”.
He explained that the system consists of an emitter and a receiver, shooting high-energy radiation through materials. It can penetrate up to 150mm of contiguous steel, effectively scanning the sides of a 40-foot container truck. The receiver provides readings related to the chemistry, relative atomic mass, and the shape and size of objects within the scanner.
“The driver is typically totally safe because these systems look for the coupling between the cab and the truck, and they only start shooting after the cab passes through. On all our conveyor belt systems, for example, we put up lead panels on both sides and then we show up on site with a Geiger counter before we commission.”
Tom Bird, Executive Chairman, ENICOR (GBR), wondered how the technology would work at scale with large loads potentially containing many different batteries? Mr Mecheri said each battery type had its own risk profile so assessments of those individual risks helped operators decide which ones could safely be processed and which ones had to be isolated and dealt with.
Mr László described how Kuusakoski Recycling has a line going through a sorter with two cameras. “One is on the line, and one is after it. The one on the line is telling the ejector there is a battery, so it shoots it out to a box with sand so we can contain it instantly, making sure it doesn’t blow up if is critically damaged.
“The other camera that’s after the line is feeding back all the time the data on how well the ejection is working. It is self-learning.” He explained the system may require manual adjustments if it misses something. “Then a person goes and reprograms it: it’s a human-machine interaction.”
