Description du projet
S’appuyer sur la robotique pour faciliter la récupération de matériaux recyclables
Les matériaux recyclables sont généralement triées manuellement dans des installations de récupération des matériaux (IRM) situées à proximité de zones urbaines denses. Les récentes évolutions en matière d’intelligence artificielle (IA) et de robotique permettent l’automatisation de plusieurs activités des IRM. Toutefois, cela s’avère souvent peu rentable et peu adapté aux grands volumes de déchets comme aux petites zones. Si les unités mobiles de valorisation des matériaux semblent apporter une solution à ce dernier point, elles ont leurs limites. En s’appuyant sur des technologies éprouvées en robotique, en IA et en analyse de données, le projet RECLAIM, financé par l’UE, entend développer une IRM automatisée mobile, adaptée à la récupération de matériaux à petite échelle. Ce projet adoptera une méthode modulaire à multirobots ou multipinceurs pour la récupération des matériaux, dans le cadre d’une approche de science citoyenne.
Objectif
Recyclable materials recovery is a key element of the circular economy and the EU Green Deal. It is typically performed manually at large scale Material Recovery Facilities (MRFs) installed close to dense urban areas. Recent advances in AI and robotics have enabled the automation of several MRF activities. However, they target large waste volumes and are not cost-effective for smaller, less accessible areas.
To accommodate the latter, portable material recovery units can be deployed nearby. Despite the increasing demand for portable units, offerings lack intelligent, automated components that could significantly increase their productivity.
To fill this gap, RECLAIM will develop a portable, robotic MRF (prMRF) tailored to small-scale material recovery. The proposal exploits well-tested technology in robotics, AI and data analytics which is improved to facilitate distributed material recovery.
RECLAIM adopts a modular multi-robot/multi-gripper approach for material recovery, based on low cost Robotic Recycling Workers (RoReWos). An AI module combines imaging in the visual and infrared domain to identify, localize and categorize recyclables. The output of this module is used by a multi-RoReWo team that implements efficient and accurate material sorting. Further, a citizen science approach will increase social sensitivity to the Green Deal. This is accomplished via a novel Recycling Data-Game that enables and encourages citizens to participate in project RTD activities by providing annotations to be used in deep learning for the re-training of the AI module.
RECLAIM developments will be implemented and repeatedly assessed in demanding, real material recovery tasks. Three different scenarios will attest its effectiveness and applicability in a broad range of locations that face material recovery challenges. This will pave the way for the prMRF market uptake and provide a major boost in making Europe zero polluting, climate-neutral, sustainable and globally competitive.
Champ scientifique
- engineering and technologyenvironmental engineeringwaste managementwaste treatment processesrecycling
- natural sciencescomputer and information sciencesartificial intelligencemachine learningdeep learning
- social sciencespolitical sciencespolitical policiescivil society
- engineering and technologyelectrical engineering, electronic engineering, information engineeringelectronic engineeringrobotics
- social scienceseconomics and businesseconomicssustainable economy
Programme(s)
Régime de financement
HORIZON-IA - HORIZON Innovation ActionsCoordinateur
70013 Irakleio
Grèce