loading page

AI-Driven Sustainable Weed Managing Mobile Robot
  • +6
  • Hadi Moeinnia,
  • Devin Armstrong,
  • Jacob Angelozzi,
  • Gavin Theys,
  • Kimia Rezaeian,
  • Yonghao Wen,
  • Devon Scott,
  • Emmanuel Sulle,
  • Woo Soo Kim
Hadi Moeinnia
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Devin Armstrong
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Jacob Angelozzi
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Gavin Theys
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Kimia Rezaeian
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Yonghao Wen
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Devon Scott
Simon Fraser University School of Mechatronic Systems Engineering
Author Profile
Emmanuel Sulle
The Aga Khan University - Tanzania
Author Profile
Woo Soo Kim
Simon Fraser University School of Mechatronic Systems Engineering

Corresponding Author:woosook@sfu.ca

Author Profile

Abstract

This study introduces a compact, autonomous mobile weed management robot designed to promote sustainable agricultural practices and enhance crop protection through effective early-stage weed management. Equipped with a laser-based system, the robot enables precise weed removal tailored to specific agricultural contexts. It employs an AI-driven image classification approach for weed detection, achieving a mean average precision (mAP) of 0.32 and a detection rate of 118 ms on a Raspberry Pi 5 platform. The robot features a two-degree-of-freedom arm for accurate laser positioning, with exposure duration dynamically adjusted based on identified weed species to minimize energy consumption and protect neighboring crops and soil. Field trials in Vancouver, Canada, and Arusha, Tanzania, demonstrated the robot’s effectiveness, achieving weed removal success rates of 97% and 96%, respectively, in a maximum of 60 seconds targeting pigweed, purslane, and nutsedge. Designed to be cost-efficient and scalable, this innovative system offers an environmentally sustainable solution for effective weed management, significantly reducing herbicide use and enhancing weed targeting precision. This research underscores the dual benefits of integrating autonomous technology into agriculture, improving productivity and sustainability while protecting crop health and ecosystems.