ARTIFICIAL INTELLIGENCE AND DATA SCIENCE IN RECYCLING & MATERIAL RECOVERY: A SYSTEMATIC LITERATURE REVIEW

Abstract

Less than 20% of plastic waste is effectively recovered globally. The potential for artificial intelligence (AI) and data science to revolutionize the recycling and material recovery systems, from mere reaction to automation and intelligent management, is unparalleled. This systematic literature review aims to explore the use of artificial intelligence and data science methods, such as deep learning, machine learning, computer vision, reinforcement learning, and natural language processing, in recycling and material recovery scenarios and summarize the existing knowledge about their applications, highlighting the main research gaps and proposing the SMART-R conceptual framework as one model that unifies the scope of intelligent material recovery. A systematic search was carried out according to PRISMA 2020 guidelines in the four popular databases: Scopus, Web of Science, IEEE Xplore, and Google Scholar, for the period of 2015-2025. Of the 87 studies that passed the screening process, 80 reported the results of their secondary analysis. Of these 87 studies that passed screening, 80 reported results of their secondary analysis. Accuracy of waste classification models based on deep learning on benchmark datasets is 85–97%. Machine learning techniques show great promise in predicting waste and detecting waste contamination. Yet there are still significant challenges in end-to-end system integration, deployment, and applicability in developing countries. The proposed SMART-R framework (Sensing, Mapping, Analysis, Routing, Tracking, Reporting) is a unifying structure for AI-enabled recycling systems. Real-world validation, edge deployment, and human-AI collaboration in sorting facilities should be emphasized in future research.