From data to decisions: How can AI and big data support decision-making in resource recovery from waste?

How can AI and big data support decision-making in resource recovery from waste in industrial sectors? This central question was at the heart of the REACT Cluster’s third webinar episode on May 9, led by RECLAIM project and supported by iBot4CRMs project, which brought together 53 participants from across various industries. From research to academia, recycling companies to technology providers, the webinar captured a diverse audience base for a lively knowledge exchange.
Moderated by Léa Zamuner, the technical communications coordinator for the International Solid Waste Association (ISWA) and the iBot4CRMs project, the session started with presentations by specialists Michalis Maniadakis from the RECLAIM project, and the Institute of Computer Science-FORTH, Petri Kannisto from the ALCHIMIA project and the Betriebsforschungsinstitut, Antonio M.Ortiz from the iBot4CRMs project and the Norwegian Research Centre, and Cesar Arenas Prado from the DARROW project and the Ceit:
RECLAIM’s Michalis presented the portable robotic Material Recovery Facility (prMRF), which is touted as the future of decentralised material recovery activities for remote areas in Europe:
- Petri introduced a decentralised Artificial Intelligence (AI) and data solution to support the green transition of Europe’s major metallurgy industries:
- Antonio explained how AI-powered robots are being developed to automate urban mining and improve the recovery of critical raw materials;
- Cesar highlighted efforts to enhance wastewater treatment by reducing energy use, emissions, and chemical consumption while improving water quality.
All the projects also introduced how they use AI for resource recovery from different waste streams and sectors, such as municipal solid waste, industrial waste, urban waste and wastewater. While each waste stream is different, all these projects had a unified goal to collect data to improve the processes and increase efficiency to ultimately make European industries more competitive.
In this regard, the panel discussion delved into key themes, such as data usage for robot control, evaluating robotic decisions, and the integration of AI and non-AI methods for data-driven decision-making. Participants also discussed data collection strategies, AI-assisted process optimisation, and the challenges of scaling up these solutions.
Following an engaging Q&A session with the audience, the experts were invited to explore potential areas of collaboration among their projects. Clear opportunities emerged between RECLAIM and iBot4CRMs, both of which employ similar robotic systems, such as vision-based component identification. Other promising topics included shared methodologies and the use of continual learning approaches in AI. Petri also called attention to the overarching goal shared by all the projects: advancing a circular economy and sustainability while safeguarding human welfare and building resilience. He underlined the importance of continuing this collaborative work to achieve that common vision.
Key takeaways from the webinar:
Data quality is the foundation for AI success
- (Michalis) emphasised that the efficacy of AI and Data systems rests on the quality of the input waste stream. For RECLAIM, industries must ensure consistent and standardised waste types (e.g., Plastic, Metal and Drinking Cartons (PMD) for a solution like prMRF to be effective.
- Poor-quality or inconsistent inputs undermine optimisation and recovery performance.
Integration of AI, Robotics, and vision systems must be seamless and profitable
- Antonio from iBot4CRMs) pointed out that for AI-driven solutions to scale, they must integrate robotics, computer vision, and AI models effectively.
- Adoption depends heavily on economic viability — if Critical Raw Materials (CRMs) recovery processes are not profitable, waste management plants won’t adopt them.
Sustainable models beyond research projects
- Petri From Alchimia said to ensure long-term impact, AI optimization software must transition from research projects to operational use through either a sustainable service model or operator-led adoption.
AI will not replace plant operators — It will empower them
- (Cesar) stressed the importance of gaining operator trust. AI tools are meant to support human decision-making, enhance energy efficiency, and simplify complex operational decisions — not eliminate human roles.