Precision agriculture
GPS-steered machines, drones and image-recognition sprayers apply seed, fertilizer and herbicide only where the field needs it.
Open in the interactive tree →Satellite guidance and yield maps spread from the 1990s. Today's systems add cameras with deep-learning models that spray only weeds, autonomous tractors and sensor-based variable-rate application. The aim is the same yield with fewer inputs.
As of October 2026
Blue River, bought by John Deere in 2017, has claimed that its camera-guided sprayer cuts herbicide use by up to 90%; this is a vendor claim, not an independent field measurement. Deere reports (company-reported, baseline not stated) that its See & Spray system covered more than five million acres in 2025 and cut non-residual herbicide use by nearly 50% on average, saving nearly 31 million gallons of herbicide mix. Deere showed a self-driving tractor for large farms in January 2022. Adoption is still uneven: fewer than 10% of farmers in France and in low- and middle-income countries use variable-rate systems, and apple-picking robots need 5-10 seconds per fruit against about one second for a person, which is too slow to pay.
Open steps
- Real-time weed detection on tractors High AI leverageBuild vision models running on tractors and sprayers at 10+ fps, segmenting weeds from crops in mixed vegetative growth for sub-meter spraying.
- Six-week yield and disease forecasts High AI leverageFuse satellite, drone, soil and weather data into per-zone yield and disease risk maps 6-8 weeks before harvest, actionable for intervention.
- Fast fruit picking via learned grasping Medium AI leverageUse 3D vision and learned grasp planning to pick delicate crops at 3-5 seconds per fruit instead of 10-20, unlocking economic viability.
Where AI could help
Medium AI leverage. Image-based spraying and yield models are field-tested, but machine cost, small farms and slow picking robots limit the rollout.
- Detect weeds and crops in real time so sprayers treat only plants that need it
- Fuse satellite, drone and soil-sensor data into per-zone fertiliser and water plans
- Speed up picking robots with better fruit detection and grasp planning
- Predict yield and disease risk early enough to act
Shown so far
- In July 2025, a University of Florida study reported that AI-guided targeted spraying cut herbicide use by 71% in tomato fields with effective weed control. source
Prerequisites
- Combine Harvester1836Yield monitors on combine harvesters started precision farming
- The tractor1917
- Earth observation by satellite1960Field-level crop monitoring and variable-rate maps use satellite imagery
- Satellite navigation (GPS)1995
- Deep Learning2012