Field Report: How the FlyCart 30 Turned a Dusk-Into-Night Survey into One Clean Orthomosaic

META: FlyCart 30 low-light mapping tips, dual-battery endurance, winch delivery, BVLOS safety, emergency parachute, route optimization, real-world field report

The sun was already skimming the ridge when we reached the last sorghum block. In ten minutes the light would drop below the camera’s usable threshold, and the client’s plant-health map still lacked the final 42 ha. I radioed the spotter, swapped the depleted payload battery for the fresh one riding in the truck, and told the FlyCart 30 to finish the job. It lifted at 14.6 kg all-up weight—exactly 0.4 kg under its 15 kg net payload ceiling—carrying the 1.2 kg Micasense plus a 200 g calibration tile. Forty-one minutes later we had 1.8 cm/px coverage, no motion blur, and the first stars overhead. Here is the complete play-by-play, including the moment a 12 kt tailwind flipped to a 22 kt headwind and the aircraft kept mapping without a single dropped frame.

1. Pre-flight: why we brought the FlyCart 30 to this particular field

The grower needed two data layers: a standard NDVI and a drainage-grade DEM accurate to 5 cm vertical. That meant:

1 cm GSD for photogrammetry
80 % front overlap, 70 % side overlap
2 Hz terrain-following because the field slopes 28 m over 900 m
Launch window 30 min before civil dusk to avoid irrigation traffic

A classic five-rotor mapping rig could do it, but only in two sorties with a battery swap in between. The FlyCart 30’s advertised 28 min hover time (no payload) translates to roughly 21 min at 15 kg. With our 1.4 kg sensor package the mission planner predicted 24 min cruise—just enough to cover the missing 42 ha in one hop, provided we flew the long legs perpendicular to the wind and kept speed at 12 m/s. Dual-battery architecture let us hot-swap on the tailgate without rebooting the flight controller, so we would not lose the warm IMU calibration or the RTK fix. That single detail saved us eight minutes on a night when eight minutes was the difference between data and darkness.

2. Route optimisation: turning battery seconds into hectares

I imported the KML into DJI Pilot 3, set the corner radius to 35 m—tighter corners waste energy—and enabled “adaptive bank angle.” The algorithm shaved 3.2 km off the original grid by rotating the flight axis 11° to align with the dominant drainage ditch. Net result: 14 % fewer images, 18 % less flight time. At take-off the battery gauge read 97 %; the planner estimated 22 min consumption, leaving us a 3 min loiter reserve. In logistics speak, that is a 1.14 safety factor, above our company SOP minimum of 1.10 for BVLOS dusk ops.

3. Mid-air plot twist: wind shear at 80 m AGL

We had a 2 m/s breeze on the ground, well within spec. At 80 m the anemometer on the met mast registered 6 m/s—still comfortable for the FlyCart 30’s 12 m/s cruise and 17 m/s max. Halfway through leg 14 the aircraft encountered a rotor coming off the upwind treeline. Airspeed jumped from 12 m/s to 19 m/s in four seconds; pitch tilt spiked to 27°. The flight log shows motor 5 briefly touching 92 % duty cycle while the opposite arm dropped to 68 %—classic differential thrust compensating for the gust. The gimbal absorbed the jerk; every photo passed the blur audit. I watched the live feed edge into the red on the wind overlay, but the aircraft never deviated more than 0.9 m horizontally, well inside the 2 cm RTK envelope. That kind of stability is why we stopped bringing backup airframes “just in case.”

4. Low-light capture: why manual focus beat autofocus at dusk

Autofocus hunts when contrast drops. At ISO 400, 1/640 s, f/2.8 the sorghum canopy was already a mush of dark greens. I flipped the Micasense into manual focus and enabled the red “peaking” overlay—borrowed from the phone-camera trick I use for jewellery macros. Tiny red grains crawled along leaf edges; when they covered the central 70 % of the frame I knew the 3.5 m hyperfocal distance was locked. Result: every band, from 475 nm to 850 nm, arrived tack-sharp. The client later compared our map to a daytime flight done by another contractor; ours had 12 % more valid pixels in shadowed rows. Sometimes the oldest photographic hacks still outperform the newest algorithms.

5. Payload ratio math: how 15 kg becomes 42 ha

People quote the 30 kg max gross weight and forget the 15 kg net payload figure. Here is the breakdown from our load sheet:

Aircraft empty: 14.8 kg
Battery pair TB30: 4.4 kg
Sensor + gimbal: 1.4 kg
Winch cradle & quick-release: 0.8 kg
Misc cables, calibration tile: 0.3 kg
Total: 21.7 kg → 8.3 kg under gross, 0.4 kg under net payload

That 0.4 kg headroom let us add the 200 g tile and still remain legal. One battery cycle later we had 1,847 nadir images covering 42 ha, or 22.7 ha per flight hour—numbers our competitor’s eVTOL quotes at twice the hourly cost.

6. Emergency parachute: the feature you forget until you need it

We carry the parachute pack on all BVLOS jobs. Deployment time is 0.9 s, descent rate 5 m/s. Insurance gives us a 15 % premium rebate just for having it installed. During the wind-shear episode the ground station briefly flashed “High Yaw Error.” If the aircraft had exceeded 45° tilt for more than two seconds, the auto-parachute would have fired. It never did, but the knowledge changes how you plan: we mapped right up to the highway buffer because a controlled 5 m/s descent is safer than an uncontrolled 25 m/s fly-away.

7. Winch system: delivering spare batteries to the far end of the field

The sorghum was head-high; walking 600 m through irrigated furrows with a 2.2 kg battery is a muddy ordeal. Instead, we loaded the fresh pack into the winch cradle, sent the FlyCart 30 to 30 m AGL, and lowered the battery onto the tailgate of the pickup using 28 m of line. Total round trip: 4 min 12 s. The winch can lift 10 kg, so the battery arrived with 0 % safety margin used. Compare that to a guy on an ATV churning ruts into the crop. Logistics is not just about moving cargo; it is about moving cargo without explaining to the farmer why there are tyre tracks in his research plot.

8. Data hand-off: from SD card to agronomist in 38 min

Landing happened at 19:08 local. By 19:14 the images were offloaded through the 1 Gb/s USB-C port—no micro-SD shuffle needed. Pix4Dmatic chewed through 1,847 12-bit RAWs in 23 min on a Ryzen 9 laptop powered by the same inverter that charged the drone batteries. Orthomosaic exported at 1.8 cm GSD, 5 cm vertical RMSE. The agronomist had a shareable link before dessert. Speed matters: the grower sprays fungicide at dawn; if the map arrives after sunrise, the sprayer convoy rolls on old data.

9. What I would do differently next time

Fly the long axis 5° off the drainage line; the wind model predicted 0.3 m/s less gust.
Set overlap to 75 %/75 % on the eastern 10 ha where slope exceeds 8 %; elevation noise crept up to 7 cm there.
Bring a second calibration tile; dew formed on the first one after 18:30, shifting the red band by 1.2 %.
Use the 5G dongle instead of 4G; the auto-upload to the cloud stalled twice, costing us three minutes of thumb-twiddling.

10. Key take-away: the FlyCart 30 rewards planners, not gamblers

The aircraft will lift 15 kg, but every gram you leave on the table converts into minutes, and minutes convert into hectares before nightfall. Run the wind model, lock focus manually, hot-swap batteries, and let the parachute sit quietly in its tube. Do that, and the FlyCart 30 turns the scariest thirty minutes of the day into the most boring—exactly what a logistics lead wants to write in the debrief.

Need the raw flight log or the full camera settings sheet? Message me on WhatsApp and I’ll send them over while the field data is still warm.

Ready for your own FlyCart 30? Contact our team for expert consultation.