Variable Rate Treatment System Uses AI to Optimize Crop Management

Deere & Company has developed an advanced method and system for determining variable rate treatments based on crop canopy characteristics. The technology applies computer vision and machine learning to analyze images of each field region, identify individual plants and their canopy characteristics, and autonomously control treatment application to precisely match crop conditions.

The Problem

Modern agricultural machinery has traditionally operated with single treatment rates across entire fields, despite significant plant-to-plant variation in size, vigor, and pest susceptibility. This uniform approach wastes resources by over-treating vigorous plants and under-treating those requiring greater attention. While precision agriculture has advanced significantly, most systems rely on fixed variable rate maps created before the season, unable to adapt to real-time crop conditions or unexpected changes.

Furthermore, traditional machine learning algorithms struggle to completely understand the complex dynamics of agricultural conditions and are often trained for primary functions other than treatment optimization. This limitation renders conventional systems unable to adapt or switch treatment approaches in response to the dynamic variations visible in the crop canopy during the growing season.

What This Invention Does

The system uses a canopy recognition model applied to images captured for each field region. The model identifies each plant in the region and the plant’s canopy structure, then determines canopy characteristics from the pixels representing the plants. These characteristics might include canopy density, height, leaf coverage, color variation, or other visual indicators of plant vigor and condition.

Based on these canopy characteristics, the system determines control signals for variable rate treatment, translating visual plant assessment into precise application amounts. The farming machine actuators then receive these control signals and autonomously apply the variable rate treatment. The system operates in real-time or near-real-time as the machine moves through the field, continuously analyzing crop conditions and adjusting treatment delivery to match each plant’s individual requirements.

Key Features

• Canopy Recognition Model. Machine learning model trained to identify plants and analyze canopy structure from field images with high accuracy.
• Real-Time Image Analysis. The system captures and processes images continuously as the farming machine operates through the field.
• Characteristic Identification. The model determines multiple canopy characteristics from image pixels, creating detailed assessment of each plant’s condition.
• Autonomous Control Signal Generation. Treatment requirements are automatically calculated from canopy characteristics without requiring manual input or pre-generated maps.
• Dynamic Variable Rate Application. Treatment mechanisms actuate in real-time based on control signals, enabling true dynamic variable rate application.

Who Is Behind It?

Deere & Company, the global leader in agricultural equipment, developed this innovation with inventors Christopher Grant Padwick, David Evans, Divya Kulkarni, and Ripudaman Singh Arora. The patent was filed on 25 August 2025, claiming priority to U.S. Application 18/821024 filed on 30 August 2024. Pizzeys Patent and Trade Mark Attorneys Pty Ltd represents the application in Australia.

Why It Matters

This patent represents a significant advancement in precision agriculture and crop management technology. By enabling real-time assessment of crop conditions and corresponding treatment adjustment, the system can substantially improve crop outcomes, reduce input waste, and enhance sustainability. Farmers can optimize insect management, disease control, and nutrient application for each plant individually rather than applying uniform treatments across diverse crop conditions.

The development of systems that understand crop variability through machine vision and respond autonomously reflects broader trends toward intelligent agriculture and aligns with global needs for more efficient, sustainable food production. The technology enables more responsive farming that adapts to actual conditions rather than relying on predetermined maps and assumptions.

Related Concepts

Precision agriculture uses data, sensors, and automation to tailor farming inputs – such as water, fertiliser, and pesticides – to the specific needs of individual crop zones or plants. Technologies like GPS field mapping, remote sensing, and variable rate application systems allow farmers to reduce waste and improve yields by treating each part of a field according to its actual conditions rather than applying uniform rates.

Machine vision in agriculture applies image processing and deep learning to analyse crop imagery in real time, enabling autonomous identification of plant health, pest damage, growth stage, and canopy structure. Combined with variable rate technology, these systems can automatically adjust chemical application rates as a farming machine moves through a field, responding dynamically to observed crop conditions.

---

AU 2025220885 was published in the Australian Official Journal of Patents on 19 March 2026 and is open for public inspection. Patent applications represent inventions that are sought to be protected and do not necessarily reflect commercially available products.