COBOL: COmmunity-Based Organized Littering

ittering is a significant issue that threatens the environment, society, and the economy. It does not only affect urban areas but can also turn suburban and rural areas into open-air dumps, causing serious damage to flora and fauna. The collection of waste entails substantial economic and social costs, and the affected areas are often protected zones of great value for the regions. Monitoring, tracking, and regularly cleaning these sites is a complex task that involves public authorities, local administrations, companies, and citizens. Previous solutions have either relied on highly technological but expensive approaches or on methods that provide limited support to waste management while depending heavily on users continuously submitting reports. Within this context, COBOL is a national project funded by the Italian Ministry of University and Research and launched in December 2023. It is coordinated by the University of Milano-Bicocca with the participation of the Polytechnic of Milan and the Gran Sasso Science Institute, and it collaborates with several Italian municipalities interested in testing the services developed. COBOL aims to define a decentralized architecture for waste management that integrates gamification to engage all stakeholders, model-driven engineering to enable the creation of flexible and executable waste management models, computer vision to identify, sort, and categorize waste, federated learning to share knowledge among different communities without violating privacy, and self-adaptation to handle unexpected events. Early results indicate that the federated learning architecture can effectively collect reports and that computer vision techniques applied to real-life datasets can support the semi-automation of litter detection.