Next generation vision guidance system launched by Garford Farm Machinery
The pioneer of the Robocrop advanced precision guidance technology for mechanical hoeing and weed control, Garford Farm Machinery has now launched the next generation, multi-modal vision guidance system during the World Agri-Tech Innovation Summit in London (Sept 30-Oct 1, 2024).
Delivering higher uptime, increased capacity and lower cost of operation, the new Robocrop AI InRow option has been developed exclusively by Garford’s project team of mechatronic, software and data engineers, to provide greater flexibility and higher resilience to light conditions and weed pressure.
Representing a leap in vision guidance capability with ‘green on green’ plant recognition, Robocrop AI uses artificial intelligence to combine colour, infrared and depth information to precisely identify crop plants and stem location.
“Green on green plant recognition is a step change from the green on brown system that uses colour only. This new system uses multiple different data streams from the camera, and is able to then identify crop plants and distinguish those from weeds or non-crop plants,” explains Garford Farm Machinery managing director Jonathan Henry.
A big step function benefit of Robocrop AI, differentiating it from other AI systems, is the multi-modal camera system. By using infrared cameras, as well as a colour camera, depth imaging, calculating vegetative index and then using data fusion techniques, the system accurately detects stem locations, even where multiple leaves overlap.
“In practice, this means we can hoe closer to the plant to achieve a higher weed control efficacy, ultimately bringing value to the user,” adds Mr Henry.
The new vision guidance system also brings new levels of resilience to varying light conditions and cloud cover, as well as weed pressure in the field.
The multi-modal camera is combined with an AI neural network, which is trained to identify and recognise the crop plant and stem location, then distinguish it from weeds and non-crop plants, using colour, as well as leaf shape and profile.
“Using multiple data streams means that, if for example due to weather conditions the operator is going out
late and the weed pressure has intensified, you can still identify the crop plants and hoe with high precision,” Mr Henry says.
Available at launch for the 2025 Garford InRow electric rotor weeders for use in salad, brassica and sugar beet crops, the first release Robocrop AI will feature single section steering systems, with single and double camera configurations.
The system features a 10.1” terminal with a Garford developed interface with live stream images and sideshift positioning. Expanding camera support and multiple sections will follow in future seasons.
Available across Garford Farm Machinery’s network of UK and international dealerships and distributors for the 2025 season.