Development and field testing of the next generation of vision-guided weeding systems

"The current crop production systems have been reliant on the wide-scale application of herbicides to control weeds. However, this approach is not sustainable due to unprecedented regulatory and environmental pressures which place new emphasis on the development of novel techniques to kill weeds. Precision hoeing equipment is a promising alternative but the current systems cannot achieve the accuracy required to completely remove the weed presence from the fields. This project will develop the next generation of robotic weeding machinery, enabling selective and accurate treatment of specific weeds.

The proposed technology is a novel combination of unique precision hoe, guided by a vision system based on multi-modal sensing and state-of-the-art machine learning for precise detection and localisation of the weeds. The system will be re-trainable by the end-user and enable rapid deployment of the system and adaptation for various plant types and environmental conditions.

The proposed weeding system will be modularised and scalable so it can be deployed both as a tractor-mounted solution building on Garford Farm Machinery's RoboCrop system and on the autonomous mobile robot Thorvald developed by Saga Robotics Ltd. An autonomous weeder can weed at lower speed and with greater precision while enabling continuous 24/7 operation. In addition, it offers a solution for farmers who grow crops in glasshouses or polytunnels where the traditional implements cannot be deployed. The proposed developments will lead to more efficient weeding equipment resulting in better management of weeds and reduced input use, bringing several economic, social and environmental benefits to food producers, sellers and society."