Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{195148,
        author = {Rashi R. Ladvikar and Prof. Aniket R. Thakur and Mrinmayee A. Khonde and Ishwari V. Munde and Vivek D. Gazalwar and Shreeya S. Lande},
        title = {SmartWaste: Real-Time Urban Waste Detection Using YOLOv8 and Multi-Agent Reinforcement Learning for Intelligent Task Allocation},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {10},
        pages = {7241-7249},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=195148},
        abstract = {Rapid urbanization has intensified the challenges of municipal solid waste management, demanding intelligent, scalable, and real-time solutions. This paper presents SmartWaste, an AI-driven urban waste management framework that integrates real-time computer vision, Multi-Agent Proximal Policy Optimization (MAPPO), and an incentive-based citizen engagement module. The vision module employs a YOLOv8n object detection model trained on a custom dataset of 2,847 annotated images across three waste classes (trash, liquid spillage, and bin overflow), constructed using Roboflow with diverse urban scene augmentation. The MAPPO-based decision layer coordinates sanitation worker assignments dynamically in response to detected waste events within a simulated urban grid environment. A gamified browser-based reporting interface supports citizen participation, while a centralized analytics dashboard enables data-driven municipal decision-making. Experimental results demonstrate that the system achieves a detection precision of 77.6% and mAP@0.5 of 75.2% on held-out test images, and MAPPO-based allocation reduces average waste response time by 51.6% and improves workforce utilization by 44.5% over static zone-based methods. These findings validate the practical effectiveness of integrating vision-based perception with multi-agent learning for smart city waste management.},
        keywords = {Smart Waste Management; Computer Vision; YOLOv8; Multi-agent reinforcement learning; MAPPO Smart Cities; Task Allocation; Citizen Participation; Urban Analytics},
        month = {March},
        }