Drone data collection: from flight to platform

A drone is a sensor that moves. It passes over the field or the industrial plant and brings back images that, once processed, become a map with numbers: crop vigor, the temperature of each point, terrain, area. The value of that data shows up when it sits next to what the fixed sensors already measure, on the same dashboard and on the same timeline.

This article walks through the whole flow, from flight plan to platform. We don't sell off-the-shelf drones: we design the onboard system when a project calls for it and we integrate data from any aircraft into the platform. So this is about the data, not the airframe.

Flight plan

Before takeoff, you define what you want to measure, and the plan follows from that:

• Area and altitude. Flying lower gives more detail per pixel (the GSD, or ground sample distance) and takes more photos to cover the same area.
• Overlap. To build a mosaic, each photo has to overlap its neighbors; 70 to 80% along the flight line and 60 to 70% sideways are typical in agricultural photogrammetry, and the processing software tells you what it needs.
• Time of day. Multispectral needs a high sun and stable sky; thermal work on crops does best mid-day, when water stress is visible, or at dawn for soil differences.
• Ground control points. Targets on the ground with known coordinates, when positional accuracy matters (area, volume, design work).

The plan becomes an automatic route. The pilot supervises rather than flies.

Sensors

• RGB. The ordinary camera. It produces the visual orthomosaic, the elevation model and plant or gap counts along the row.
• Multispectral. Separate bands: red, red edge, near infrared (NIR). They let you compute vegetation indices. The best known is NDVI = (NIR − Red) / (NIR + Red), which ranges from −1 to 1 and rises with green biomass.
• Thermal. A long-wave infrared radiometer. It measures surface temperature: a stressed leaf heats up, a wet patch cools down, a bad electrical connection runs hot.

What comes out of a flight

Photogrammetric processing stitches hundreds of photos together and produces:

• Orthomosaic: a single corrected, georeferenced image.
• Digital surface and terrain models, for relief, drainage and stockpile volumes.
• Index maps (NDVI and others), per pixel and summarized by field, by zone or by pivot span.
• Temperature points, with coordinates, for inspecting solar panels, substations, piping and roofs.
• Detections: planting gaps, weeds, thermal anomalies, tagged by position.

These are all large files. What the platform needs is not the whole image; it is the numbers per area and per point, with date, time and coordinates.

How it gets into the platform

The platform receives data through an API, from any source. For drones, the flow we integrate is:

1. The processing software exports the summaries: mean and minimum index per zone, a list of anomalies with coordinates and temperature, measured area.
2. A connector sends them to the API, linking each number to an entity already registered: the field, the pivot, the panel string, the piece of equipment.
3. The platform stores the value with the flight time and puts it on the same time series as the fixed sensors: weather station, soil moisture, pivot flow, engine hours, vibration sensor.

The result is that the field's NDVI shows up next to that week's ETo and applied irrigation depth. The thermal anomaly on a panel shows up next to the drop in string current. The drone confirms what the fixed sensor flagged, or points to what the fixed sensor can't see.

Rules for flying

Drone rules depend on the country, and many aviation authorities classify operations by risk. In the US, commercial flights with small drones fall under FAA Part 107; in the European Union, EASA sorts operations into open, specific and certified categories, and most agricultural mapping and inspection is planned to fit the open one, which carries the lightest requirements. Flying beyond visual line of sight, close to people or near airports usually moves an operation into a stricter regime that needs prior authorization.

Depending on where you fly, the usual checklist includes aircraft registration, pilot training or a knowledge test, airspace authorization, radio equipment approved for that market and liability insurance for commercial work. These rules change from time to time: before hiring a service or flying, check the current rules of your aviation authority.

What to do now

1. Define the question the flight answers: planting gaps, water stress, a hot panel, stockpile volume. The sensor and the plan follow from that.
2. Check whether the operation fits the lowest-risk category in your country's rules. If it does, the regulatory cost is low.
3. Decide where the data will live. A map on the pilot's laptop gets lost; a number on the platform, next to the fixed sensors, becomes history.

If you'd like to discuss a project, talk to one of our engineers.

---

Sources: Rouse et al. (1974), definition of NDVI.