The comments were made during a webinar hosted by MRW on 23 September that was held in association with technology group EverestLabs.
MRW editor Neil Merrett hosted the webinar as a means to discuss some of the core considerations linked to system adoption and how to look at effectively implementing AI-tools for managing tasks such as sorting and waste stream analysis. The event heard how a growing number of tools and services are being developed across the sector with a range of companies and organisations looking at how they might configure their operations to make use of these systems in line with specific needs.
You can view a full video of the event in the player below to hear some of views of speakers working across the sector to share views about some of the practical consideraitons that the adoption of AI-tools might mean for recycling operations:
Ambitions for AI integration
The event was divided into two distinct sections, with the first being a panel discussion hearing views on the impact of both existing and future developments around digital tools in line with addressing some of the pressing challenges concerning material loss and recycling efficiency.
For the opening session, the webinar heard from EverestLabs founder and chief strategy officer JD Ambati, and WRAP senior specialist Adam Herriott. The speakers discussed some of the broader developments around industry adoption and ambitions for AI-integration.
Adam Herriott opened the event by defining some of the efforts underway across the sector to look at how AI tools might inform a ‘better’ recycling process.
He argued that there was an increasing range of functions for AI tools to support arounf recycling and resource management that extended behind opportunities for automation. This included opportunities for material stream analysis and informing supply chain developments.
He said: “There’s lots of stuff around data that I think is quite exciting looking at better packaging design.”
Herriott said this potential could apply to granting a better understanding to operators around the impacts of recycling processes on packaging as products pass through systems. It might also help inform planning around the shape, design and labelling of certain products with the aim to limit material loss, he said.
Another potential impact discussed during the session was based on helping to manage more challenging materials and waste streams at MRF sites.
Rethinking the MRF site
Herriott, who has previously worked as an MRF recycling manager before joining WRAP, said that the emergence of a different AI technologies and services could also serve to help some operators meet their obligations concerning manual sampling in Ine with regulatory requirements.
He argued that effective implementation should notably focus on looking at addressing specific issues concerning understanding or compiling data, as opposed to looking at more blanket adoption.
The webinar noted that the Environment Agency has no barrier on using AI technologies for materials facility sampling. An update from the regulator last year said that operators were required to be able to demonstrate their sampling methodology and the effectiveness of its results, regardless of the technology approach used.
One important consideraitons identified during the discussion for any plan for AI use in the range of technologies and tools on the market with applications for recycling and resource management.
This required some strategic thinking about the specific needs or challenges facing operators, added Herriott.
He said; “I think, for me, the important thing it to think, ‘what challenge am I trying to fix, and can AI help it?’
He argued that this was a preferable approach to operators moving to a blanket adoption mode, without thinking about specific issues or needs they have.
Herriott said: “I think we need to make sure you are clear on what you need to fix and how it can help.”
Some of this potential for change, beyond looking at automation within recycling plants, was the potential to use operational data from sites and across supply chains to understand how certain materials or products are processed.
Herriott said: “[So] what’s happening to material, or what material isn’t going to the place we thought is should be, and why is that? Is that because it’s a shape, or a colour or design? Is it because there is a wrap around it.”
“It could be, as well, understanding how much of a material is going into a recycling system, versus how much is not making it to the recycling system. So we understand then potentially what consumer behaviour could be like and if there is any education we need to do.”
Data focus
JD Ambati said that he viewed some of the more noteworthy impacts of AI tools for recycling in being able to offer insights and guidance derived from complex operational data. This could help capture the highest value of materials, while lowering costs. Another potential benefit was to help bolster the skills of existing operators, said Ambati.
He said: “it’s not about replacing people, it’s not about running dark plants without any people. It’s about how do you take information in real-time and optimise your processes and operations to increase recovery.”
Ambati said that another key point of promotion for many AI companies was in offering a means to limit what amounts to the potential billions of dollars in annual lost revenue due to materials that are suitable for recycling being consigned to landfill or energy recovery and incineration operations.
Ambati was asked during the session about how both current and additional generations of AI tools could impact and influence the design and operation of MRFs and recycling sites over a period of eight-to-ten years.
He said: “The new plants that are coming online will have AI built into the development phase itself. They will have AI cameras, hardware and software throughout the process from the start to finish.”
This prediction was based on work by companies such as EverestLabs as a result of the transformation of existing plants in markets such as North America and the return on investment being provided to companies.
Ambati added: “Now It’s almost a given that they will be used in new plants with AI included in the project from day one.”
Adam Herriott said the UK was seeing a number of projects integrating Ai tools and systems into MRFs and other forms of recycling facilities. These applications included examples where an AI-system had been introduced to existing operations, or included within the original design plans.
He cited one such example as a site operated by Sherbourne Recycling in Coventry that was owned by eight local authorities in the west midlands.
Herriott said: “I think there is about five people working at that plant that are all engineers making sure everything is ticking along rather than anything else.”
The discussion did touch on some of the skills issues and prospects that development might pose from a workforce and skills perspective.
Herriott added that there would also be potential to help a range of staff with making decisions on specific operations and how to ensure effective maintenance.
He added that case studies already existed where AI tools and systems were helping identify issues around efficiency and performance degradation before they can result in more significant disruption.
The scope of some systems to identify operational issues was likely to be particularly important development, said Herriott.
He added: “I know waste managers particularly like that, because it means they can keep the recycling end product nice and clean and of a high quality. They’ve got the evidence to show what is in those bales to show to recyclers.”
The webinar heard that the data that can be produced from systems can underpin and highlight the specific value and components of recycled materials.
Some of the potential implications to integrate AI tools went beyond transforming MRF operations in regards to sorting but also influencing how recyclers and packaging suppliers can demonstrate compliance with food-grade materials criteria compliance.
Herriott noted that there were cases of packagers using AI systems to demonstrate and identify how certain recovered materials have been used for specific functions such as confectionary wrapping.
The potential applications of these tools could then help underpin efforts and supply chain monitoring to try and keep more food grade materials in circulation, he said.
Herriott also cited more recent industry reaction and interest around the potential of using AI-led processes and tools to remove more challenging materials and items such as vapes. Vapes are a particular concern across the industry over the fire risks posed when incorrectly disposed in residual waste or recycling.
AI tools on the market could serve to help identify and remove these items from collections before reaching plant areas where they are at risk of being damaged or pierced that can then lead to dangerous fires.
Herriott said: “So I think that is going to be a really exciting way that we can use AI in the coming years. Because we know that for all the bans on sales of single use vapes and other things that come in, there are still a significant problem.”
“It only takes one to make its way into a plant to cause significant damage.”
Technical workshop
JD Ambati returned for the second session of the webinar that served as a technical worship to talk about how EverestLabs has been working with companies in North America and Europe to look at integrating AI tools into operations.
This considered some of the developments both linked to robotics developments around sorting and materials management, as well as in offering more detailed guidance and assessments of potential opportunities to enhance efficiency and offer predictive maintenance.
The discussion also looked at implications for using the technology beyond the automation of certain process, and building understanding how technologies such as AI cameras and robotic Orting might support on-site resource management.
Ambati said that the company was working with partners in environments such as casinos, campuses and large-scale retail sites such as shopping centres to look at how these operations are being managed. EverestLabs has also put forward proposals for potential work at sites including Heathrow Airport.
Ambati said: “Anywhere there is on-site sorting, where there are humans touching recyclables and trash, we can put in place a robot to automatically run these operations.”
He argued that the existing human workforce and expertise could instead be focused on managing and monitoring these systems to ensure effective waste operations.
The workshop also allowed the audience to put their own questions to the company about how considerations around the energy requirements of AI factored into adoption planning, as well as how the technologies might assist with removing dangerous or hazardous materials from collections.
This story was originally published by our sister title, Materials Recycling World.