WeedXtract is a personal project focused on developing cost efective solutions for autonomous precision weeding robots. The goal is to test, validate, and refine new concepts that improve the efficiency and accuracy of preccision weeding robots in agriculture by builing a robot prototype using them.
Overview
YouTube video embeds require cookies.
RTK-GPS + Camera Row Guidance for centimeter-level accuracy.
NVIDIA Jetson Orin Nano handles localized real-time edge processing.
4 kWh LFP Battery Pack
Main developments
Completely newly developed and in many regards superior controll algorithm and weed targeting approach.
YOLOv11 draws bounding boxes around target crops (e.g., maize) in real time.
A conventional pixel-analysis algorithm maps every patch of organic green material.
Total vegetation minus target crop coordinates leaves explicit weed coordinates.
A heated contact head drops onto each weed coordinate, frying their proteins. This is a new approach that is safer and cheaper than lasers.
Previous developments
Supported by
