AI-powered self-learning robots for high-performance waste valorization and critical raw materials recovery

Project Information

iBot4CRMs

Grant agreement ID: 101189783

DOI

10.3030/101189783

EC signature date 5 November 2024

Start date 1 December 2024

End date 30 November 2028

Funded under

• Digital, Industry and Space

Total cost

€ 9 558 276,25

EU contribution

€ 8 029 445,13

Coordinated by

NORCE RESEARCH AS
Norway

Periodic Reporting for period 1 - iBot4CRMs (AI-powered self-learning robots for high-performance waste valorization and critical raw materials recovery)

Reporting period: 2024-12-01 to 2025-11-30

Summary of the context and overall objectives of the project

Europe is facing increasing pressure to secure the materials it needs for the green and digital transitions. Many everyday technologies, from electric vehicles to wind turbines and electronics, depend on Critical Raw Materials (CRMs) which are scarce, difficult to extract, and largely imported from outside Europe. At the same time, Europe produces growing amounts of complex waste, including Waste Electrical and Electronic Equipment (WEEE), End of Life Vehicles (ELVs), and diverse urban waste streams. Only a fraction of the valuable materials embedded in these products is currently recovered.

The iBot4CRMs project addresses this challenge by developing new ways to recover CRMs more efficiently and safely in line with the EU Critical Raw Materials Act. The project brings together advanced robotics, artificial intelligence (AI), sensing technologies, and digital twins to improve dismantling, sorting, and recycling processes across Europe. By combining these technological advances with insights from social sciences and humanities (SSH), the project ensures that the solutions are accepted by users, ethically sound, and aligned with societal expectations.

iBot4CRMs follows a clear pathway to impact. Its overall objectives are:

- Develop an integrated platform for waste handling: Combining AI, data analytics, simulation, and robotics. This system will be tested at four pilot sites dealing with ELVs, WEEE, metal scrap, and urban waste.
- Enable advanced recognition of valuable and hazardous materials: Through high‑resolution cameras, spectral sensors, and other sensing tools that are being combined with AI models to automatically detect and locate CRMs and hazardous components in mixed waste streams.
- Boost robotic manipulation in demanding recycling environments: The project develops robotic systems capable of dismantling complex products and extracting valuable components. These robots collaborate with human workers, supported by augmented‑reality guidance and safety protocols.
- Integrate social sciences and humanities into technology design: Following a human‑centric approach. SSH experts ensure that user needs, ethical principles, safety, fairness, and trust are embedded in the design of the sensing, AI, and robotic systems.
- Validate solutions in real industrial conditions: Four European pilot sites, located in Turkey, Greece, Spain, and Portugal, will demonstrate the technologies in real operations.
- Strengthen Europe’s resilience and circular economy: By improving CRM recovery rates and supporting new business models, the project aims to reduce Europe’s dependency on external suppliers, stimulate innovation, and foster new opportunities for industry, SMEs, and start‑ups working in recycling and digital technologies.

Expected impact:
iBot4CRMs is expected to make a significant contribution to European climate, circular economy, and industrial objectives, including the European Green Deal and the Critical Raw Materials Act. By improving the sorting and recovery of valuable materials, the project supports a more sustainable and resource efficient society.
The technologies developed are expected to:
• Increase CRM recovery rates by improving detection and dismantling performance,
• Enhance worker safety by reducing exposure to hazardous components,
• Strengthen European industry through digitalisation and automation, and
• Promote new markets for recycling technologies and services.
Beyond technological impact, the project involves strong stakeholder engagement, ethical oversight, and open science practices, ensuring transparent communication and wide dissemination of results.

Work performed from the beginning of the project to the end of the period covered by the report and main results achieved so far

During Period 1, the iBot4CRMs project achieved substantial scientific and technical progress toward developing an integrated AI‑powered robotic platform for recovering critical raw materials (CRMs) from complex waste streams. The consortium established the project’s architectural backbone by defining a multi‑layer system integrating perception, robotics, data management, simulation tools, and middleware, and implemented a first containerised prototype using Kubernetes, FIWARE, Kubeflow, and KServe to enable data interoperability and end‑to‑end AI workflows. Significant advances were made in sensing and perception, including the design and laboratory testing of inductive sensors, hyperspectral technologies, profilometers, and advanced imaging tools for CRM detection across WEEE, ELV‑related scrap, plastics, and slag. Partners pre-trained lifelong learning algorithms using publicly available datasets and created hybrid physics‑ and data‑driven process models as a foundation for future optimisation. Progress in robotics included the development of fastener perception pipelines, automated calibration procedures, visual servoing frameworks, specialised grippers, and replicated conveyor and dismantling setups that enabled early testing of manipulation and sorting tasks. Across all four pilot sites, detailed scenario definitions, KPIs, process maps, installation requirements and baseline datasets were produced, ensuring readiness for pilot deployment. Early integration activities connected perception, AI, and robotic systems through shared interfaces and communication protocols, while initial safety‑oriented human-robot collaboration concepts were developed. Collectively, these achievements establish a strong technical foundation for large‑scale validation and TRL advancement in the next project period.

Progress beyond the state of the art and expected potential impact (including the socio-economic impact and the wider societal implications of the project so far)

During the first period, iBot4CRMs progressed on diverse advances beyond the state of the art towards a unified platform for the recovery of critical raw materials (CRMs). A key breakthrough was the creation of an interoperable multi‑layer architecture combining Digital Twins, machine‑learning pipelines, and scalable middleware, enabling continuous optimisation across perception, manipulation, and process control. The project also advanced sensing and material‑perception capabilities through the development of customised inductive, hyperspectral and other machine vision solutions, along with early lifelong‑learning algorithms capable of adapting to heterogeneous, non‑stationary waste streams. These advances move beyond current industrial practice by enabling on‑line, fine‑grained characterisation of complex waste fractions and providing a foundation for autonomy in sorting and dismantling tasks. Robotic innovations included fastener‑detection pipelines with 2D/3D reprojection, automated calibration procedures, visual‑servoing methods, and specialised grippers tailored to highly variable WEEE and ELV materials. The combination of perception‑driven manipulation and model‑based optimisation represents a notable step towards highly flexible, semi‑autonomous waste‑valorisation robotics. Moreover, hybrid models for multi‑stage sorting processes, are under development integrating physics‑based simulation with real sensor data, to set the ground for self‑configuring sorting lines capable of dynamically adjusting parameters to maximise throughput and CRM recovery. Finally, the project established a comprehensive ELOS (Ethical, Legal, Organisational and Societal) framework that goes beyond technical state‑of‑the‑art by embedding trustworthy‑AI principles, human‑robot cooperation models, and early‑stage risk governance into system design.