Smart robotic system guided with YOLOv5 based machine learning framework for efficient herbicide usage in rice (Oryza sativa L.) under precision agriculture

Mohanty, Tirthankar, Pattanaik, Priyabrata, Dash, Subhaprada, Tripathy, Hara Prasada ORCID: https://orcid.org/0000-0003-0129-2642 and Holderbaum, William ORCID: https://orcid.org/0000-0002-1677-9624 (2025) Smart robotic system guided with YOLOv5 based machine learning framework for efficient herbicide usage in rice (Oryza sativa L.) under precision agriculture. Computers and Electronics in Agriculture, 231. 110032. ISSN 0168-1699

Preview

Published Version Available under License Creative Commons Attribution. Download (15MB) | Preview

Abstract

Conventional weed control methods, reliant on machinery and/or herbicide application, often incurred substantial expenses and yielded imprecise results. An innovative specialised weed control robotic method for accurate and minimal herbicide use is proposed to tackle these issues. Implementing robotic herbicide spraying, weed removal, and incorporation mechanisms along with the image recognition algorithm were introduced, leveraging intelligent automation to reduce costs and environmental hazards. Through image processing, weeds were pointed out and targeted for control in the rice field. A YOLOv5 machine learning framework underwent training using relevant datasets to facilitate precise weed management. The AI-driven robotic system, incorporating advanced image recognition capabilities, exhibited remarkable precision and swiftness, outperforming much better than manual labour in weed removal. This advancement in weed control technology helps farmers to optimise crop productivity, bolster food output, and address the ecological consequences linked with various chemicals; efforts were made to develop a prototype robotic system, which was subsequently built and evaluated in authentic agricultural settings. Experiments were carried out at the Agricultural Farm of SOA University, Binjhagiri, Bhubaneswar, Odisha, India, in a rice field, demonstrating the remarkable accuracy of the robotic system, with a minimal 2% variance from the actual weed quantities. This research highlights the promise of AI-powered weed management solutions in rice cultivation, offering economical and accurate weed detection and elimination functionalities. The robot demonstrates a superior weed control rate of 95%. In addition, the system's performance in incorporating the weeds is at a rate of 90%. It also serves as a blueprint for integrating AI into contemporary agriculture, steering the sector toward a more eco-conscious and economically sustainable future. The AI-driven solution for weed management revolutionises farming practices, equipping farmers with the tools for bountiful yields, increased economic viability, and a commitment to environmental stewardship. This underscores the imperative to prioritise scaling this innovative approach within both industrial and commercial agricultural sectors.