Efficient waste collection remains a critical challenge in rapidly urbanizing cities, where conventional fixed-route systems often result in redundant trips and overflow incidents. This study proposes an integrated framework that combines IoT-based fill-level sensing, short-term prediction using long short-term memory (LSTM) networks, and route optimization through a capacity-constrained traveling salesman problem (TSP) model solved by genetic algorithms. A three-month pilot was conducted across 150 collection points in a high-density district, with two experimental groups using the AI-IoT framework and one control group operating under conventional schedules. The results demonstrated that the experimental districts achieved a 24.8% reduction in redundant collection trips and a 30.2% decrease in overflow incidents, both statistically significant at p < 0.05. Furthermore, the LSTM prediction module reached a coefficient of determination of R² = 0.92 with a mean absolute error of 0.07, outperforming baseline regression methods. These findings confirm that integrating predictive analytics and IoT-enabled monitoring within a closed feedback loop can deliver measurable environmental and operational benefits. The framework not only enhances immediate efficiency but also provides a scalable model for sustainable urban waste management.
Clarke et al. (Fri,) studied this question.