A comprehensive review of AI-powered waste segregation: Advancing the vision of smart Saudi Cities

Abstract

• This study systematically reviews recent advances (2020-2025) in AI-powered waste segregation systems-covering computer vision, IoT-enabled smart bins, robotic automation, and blockchain frameworks-highlighting how these technologies can revolutionize waste management within Saudi Arabia’s Vision 2030 framework. • The review uniquely emphasizes the Saudi Arabian context, identifying challenges such as data localization, cultural barriers to household segregation, and institutional integration. It proposes actionable pathways for implementing AI-driven circular economy practices aligned with the Kingdom’s sustainability and smart city goals. • A detailed comparison of AI models (CNNs, transformers, hybrid learning methods) and benchmark datasets (TrashNet, Kaggle, WaDaBa, etc.) reveals that 80-85% of models achieved accuracies above 90%, yet real-world scalability and Saudi-specific datasets remain underexplored. • The paper identifies future research priorities, including federated and edge learning for privacy-aware waste systems, explainable AI for governance and transparency, and the use of digital twins and generative models to simulate waste segregation and energy recovery systems in smart cities. The growth in population, urbanization, and lifestyle shifts has led to a substantial increase in waste generation in recent years, raising serious concerns about environmental degradation, sanitation, and public health. In developing countries, industrial development and population growth have made solid waste management more challenging. To address environmental, health, and economic challenges posed by conventional waste management techniques such as landfilling and manual sorting, there is a strong need to adopt new technologies. This review examines existing waste segregation and management systems that employ Artificial Intelligence (AI), computer vision (CV), the Internet of Things (IoT), and blockchain technology, with a focus on their potential for use in Saudi Arabia's smart cities. Furthermore, this review summarizes comparative literature on the accuracy of AI models for waste management strategies worldwide. In particular, the survey covers literature from 2020 to 2025 on AI-based classification models, the automation of smart bins, robotic segregation, and cybersecurity frameworks to facilitate efficient operations. Moreover, model generalizability, data accessibility, and the lack of AI studies on regional datasets are highlighted, along with other challenges despite these advancements. Additionally, a variety of available Saudi data and policy perspectives are also included to assist future researchers and academics. To advance sustainable, AI-based waste management in accordance with Saudi Arabia’s Vision 2030, the research identifies key research gaps. It also suggests potential future solutions to explore, including explainable AI, federated learning, and digital twins. A conceptual overview illustrating the pathway from AI-powered waste segregation to energy recovery and a circular economy, aligning with the goals for smart cities in Saudi Arabia.