Design and Construction of an Intelligent Herbicide Spraying Robot
Article PDF

Keywords

Autonomous, Machine learning, Robot, Computer vision

How to Cite

Musa, M. I., & Abdulsalam, N. (2025). Design and Construction of an Intelligent Herbicide Spraying Robot. Global Journal of Artificial Intelligence and Technology Development , 3(2), 1-15. https://www.openjournals.ijaar.org/index.php/gjaitd/article/view/1383

Abstract

The increasing demand for sustainable agricultural practices has driven the development of autonomous solutions aimed at enhancing productivity, minimizing environmental impact, and reducing human exposure to hazardous chemicals. This research presents the design and implementation of an intelligent, solar-powered robotic system for targeted herbicide spraying. The system integrates a Raspberry Pi 3 Model B as its central controller, ultrasonic sensors for liquid level detection, and a real-time image processing framework developed using Python and OpenCV. A Bluetooth-enabled Android application was developed to allow remote operation, including navigation, lighting, and pump control. The image classification module employs a ten-layer convolutional neural network, trained on a curated dataset consisting of healthy crop images (maize, beans, potatoes, and groundnuts) and weeds, to accurately distinguish between target crops and invasive species. Field tests demonstrated a recognition accuracy of 98%, with a system uptime of 96% over a seven-day continuous operation. The robot achieved responsive herbicide application upon weed detection, thereby conserving chemical usage and enhancing safety. Voltage regulation within the power supply modules remained within acceptable thresholds, confirming electrical stability. The system’s design emphasizes affordability, precision, and ease of use, particularly for smallholder farmers. Although currently limited to mono-cropping environments and binary classification, the platform provides a solid foundation for future expansion into multi-crop and multi-weed detection capabilities. This work contributes meaningfully to the ongoing transformation of conventional farming into smart, data-driven agricultural systems.

Article PDF
Creative Commons License

This work is licensed under a Creative Commons Attribution-NonCommercial 4.0 International License.

Copyright (c) 2025 Global Journal of Artificial Intelligence and Technology Development

492 333