Ariasolar survey.
An autonomous photogrammetry pipeline for utility-scale solar farms. PX4 + custom mission planner, gimbal-stabilised RGB+thermal capture, RTK-pinned waypoints — 240ha surveyed in 41 minutes.
240 hectares.
A full day's work. Per crew.
Utility-scale solar operators were flying manual drone surveys with consumer hardware — one pilot, one battery swap per 20ha, no RTK correction, no thermal layer. A site inspection took a full crew day per 240ha block. Mosaics came back misaligned; hot-spot detection required a second pass.
The brief: a fully autonomous pipeline — plan a mission once, fly it repeatably, capture RTK-accurate RGB and thermal simultaneously, stitch georeferenced orthomosaics on-site, and surface anomalies before the crew packs up.
Mission planned
once. Flown forever.
Autonomous Mission Stack
We built a custom mission planner on top of PX4 + MAVSDK that ingests a KML site boundary, auto-generates a lawnmower grid at configurable altitude and overlap, calculates wind-adjusted waypoints, and uploads to the flight controller. RTK correction from a base station gives 2.3cm horizontal accuracy — no GCPs required.
The gimbal carries a dual-sensor payload: 20MP RGB for photogrammetry and a radiometric thermal camera for hot-spot detection. Both fire on the same trigger signal so every frame is spatially co-registered.
On-Site Processing
A ruggedised edge box running ROS 2 + OpenDroneMap processes imagery during the flight — by the time the drone lands, a georeferenced orthomosaic is ready. A Python post-processor runs anomaly detection on the thermal layer and overlays flagged panels on the RGB mosaic.
The operator sees a heatmap of suspect panels on a web dashboard before they've packed the ground station. No cloud upload required; everything runs on-site.
per flight.
a ground crew.
accuracy.
stitching steps.
Plan it, fly it,
read the result.
Mission planner, live telemetry, orthomosaic viewer, and thermal anomaly overlay — the full survey pipeline in one tool.
KML boundary → auto-generated lawnmower grid with RTK waypoints.
Live position, altitude, battery, and gimbal status mid-flight.
Georeferenced orthomosaic stitched on-site before landing.
Suspect panels flagged on the thermal layer before pack-up.
Panel coordinates and severity scores exported for maintenance.