Weed Map Application

The Weed Map application processes raw drone imagery to generate precision spot-spraying missions for agricultural drones and tractors. The core engine utilizes ExG (Excess Green) index analysis to accurately detect weed distribution across orthophotomaps.

This week, we completed full-cycle route generation testing. We used a heavy agricultural dataset to stress-test the offline edge-computing capabilities of the Altacoro Scout hardware.

DATASET SPECIFICATIONS

• Area covered: 6.54 ha (21.25 acres)

• Dataset size: 3.5 GB

• Ground Sampling Distance (GSD): 1 cm

• Hardware: DJI Mavic 3 Multispectral (RGB)

• Acquisition: 1 flight, grid flight plan

THE WORKFLOW

1. Data Acquisition: A scout drone completes the field inspection, and the operator retrieves its SD card.

2. Initialization: The operator powers on the Altacoro Scout tablet and launches the Weed Map application.

   
   

   

   3. Image Ingestion: The SD card is inserted, and with a single tap, the tablet begins processing the raw drone images entirely offline.

   

   4. Map Generation: Within minutes, a high-resolution orthophotomap is rendered on the screen.

   

   5. Boundary Definition: The operator taps on the map to define the precise field boundaries.

   

   5. Parameter Configuration: The operator adjusts the core parameters for mission calculation:

      • Green Sensitivity (ExG). This parameter controls the chlorophyll filtration threshold using the Excess Green (ExG) index. It helps the algorithm distinguish living green plants from soil, dead crop residue, rocks, or shadows. The slider allows the operator to adapt the detection to different lighting conditions (sunny/overcast) or specific crop characteristics.

      • Ignore Objects Smaller Than. A spatial noise reduction filter. Drone imagery can contain small artifacts, lens flares, moss, or random pixels that the algorithm might misidentify as weeds. Setting this limit (e.g., 50 cm²) filters out tiny anomalies, preventing the sprayer from wasting chemicals on false positives.

      • Spray Margin (Buffer). Creates a safety margin around the detected weed contour. Agricultural sprayers have a mechanical latency (a delay between the command signal and the actual valve opening) while moving at speed. Adding a spatial buffer ensures the target is fully covered, compensating for these mechanical delays and slight GPS drift.

      • Minimum Grid Size. Defines the spatial resolution of the final prescription map exported to the tractor's onboard terminal. This size must precisely match the physical width of a single nozzle section on the sprayer boom. This allows the tractor to know exactly which section to toggle on/off when passing over a specific grid cell.

   6. Path Calculation: The operator presses the "Calculate Weeds" button to generate the optimal spot-spraying route.

   

   7. Review & Refine: The generated mission is displayed on the map. If adjustments are needed, the operator can fine-tune the parameters and instantly recalculate the path.

      

   8. Export & Execution: Once the path is validated, the mission is exported to an SD card, which is then transferred to the sprayer drone or tractor for immediate field execution.