ASABE Technical Library - Abstract

Click on “Download PDF” for the PDF version or on the title for the HTML version.
If you are not an ASABE member or if your employer has not arranged for access to the full-text, view access and subscription options

AIoT-Driven Edge Computing System for Real-Time Optimization and Environmental Control in Poultry Farm Management

Published by the American Society of Agricultural and Biological Engineers, St. Joseph, Michigan www.asabe.org

Citation:  2026 ASABE Annual International Meeting  2600823.(doi:10.13031/aim.202600823)
Authors:   Md. Rasel Al Mamun, Muhammad Ashik-E-Rabbani, Md. Samiul Basir , Md. Bashirul Islam
Keywords:   AIoT, edge computing, environmental control, E-GRU, Random Forest, precision livestock farming.

Abstract. The Fourth Industrial Revolution (4IR) has accelerated the adoption of Artificial Intelligence (AI), the Internet of Things (IoT), and edge computing in agriculture, enabling real-time, data-driven environmental management. Despite these advances, many poultry farms continue to rely on manual monitoring and rudimentary control strategies, which are inadequate for maintaining stable microclimatic conditions in high-density production systems. This study presents the design and evaluation of an AIoT-based edge computing framework for intelligent environmental monitoring and control in poultry farms. The proposed system integrates an Arduino-based sensor substation with a Raspberry Pi 5 edge controller to continuously monitor key environmental parameters, including temperature, relative humidity, carbon dioxide (CO₂), ammonia (NH₃), carbon monoxide (CO), and airborne dust. Actuators are controlled through a relay interface using AI-driven decision logic. An Enhanced Gated Recurrent Unit (E-GRU) model is employed to forecast temperature and humidity 30 minutes ahead, while a Random Forest classifier determines optimal control actions like IDLE, HEAT, VENTILATE_LIGHT, VENTILATE_STRONG, and GAS_EMERGENCY-based on real-time sensor data and predictive outputs. Laboratory validation using 48 paired observations demonstrated strong agreement between DHT22 sensor readings and reference instruments, with a mean absolute error (MAE) of 0.14 °C and an R² of 0.998 for temperature measurement. The E-GRU model achieved predictive performance of MAE ≈ 0.47 °C (R² ≈ 0.84) for temperature and MAE ≈ 2.28% RH (R² ≈ 0.77) for humidity. The Random Forest classifier attained 99.5% test accuracy with a macro F1-score of approximately 0.99, indicating robust control decision performance. These results demonstrate the effectiveness of the proposed AIoT-edge computing system for intelligent, real-time environmental control in poultry farming, with future field-scale validation recommended to assess performance under commercial operating conditions..

(Download PDF)    (Export to EndNotes)