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.