Sumiyajid Ulziibayar,Supervisor 1:Myagmarjav. BSupervisor 2: Narangarav.T2026-11-062025-11-14Ulziibayar, S. (2025). Automated trash sorting system [Bachelor's thesis, German-Mongolian Institute for Resources and Technology].Student ID Number: B2000505https://gmitlibrary.net/handle/123456789/235The proper sorting of household waste remains a significant challenge globally, often hampered by complex logistics and inconsistent public participation. This thesis presents the design and implementation of a cost-effective, automated trash sorting system specifically engineered to address the issue locally, in Ulaanbaatar city, at the request of Nalaikh District. The system aims to actively motivate and train the population, particularly children in educational settings, to sort recyclables responsibly. The initial project scope, which involved sorting six distinct waste types using multiple sensor modalities, was narrowed due to practical and economic constraints, particularly the difficulties associated with safely storing glass and organic waste within a compact unit. We have decided to use an integrated embedded system built around a Raspberry Pi and an ESP32 microcontroller. This system utilizes a camera-based machine learning model for classification, categorizing inputs as Plastic Bottles, Metal Cans, or Others (namely, paper, cardboard, glass, and organic materials), combined with weight sensor data. The mechanical separation is handled by a combination of a stepper motor and/or a servo motor integration to make the unit more scalable. One of the key elements of the project is the development of a server-based tracking and reward system, which leverages the motivation of group competition to promote consistent usage and habit formation among its target demographic. The system achieved a classification accuracy exceeding 90% in testing, demonstrating that a highly focused, integrated engineering approach can provide a scalable and financially viable tool for improving localized recycling compliance.enWaste sortingRecyclingMachine learningConvolutional neural networkRaspberry PiESP32Internet of ThingsMongoliaThesisBachelor's thesis submitted to the Faculty of Electrical and Mechanical Engineering, German-Mongolian Institute for Resources and Technology. First Supervisor: Myagmarjav B. Second Supervisor: Narangarav T.