Dr. House: Everybody lies

“Everybody lies” - in the UAV world that means “don’t trust the sensors”. But some sensors are much more prone to talk nonsense.

Long story short, in September-October, we did many drone flights over water - with magnetometers and echo-sounders. All flights were on very low altitude under the control of a terrain-following system with a radar altimeter. My colleague Kri?jānis Linkevi?s realized that this was a good opportunity to analyze the drift of the barometric altimeter – the drone’s altitude on the survey lines was more or less constant over the surface, and the surface was almost perfectly flat.

Below are samples of data from barometric altimeters compared with a radar altimeter.

#1. DJI M600 Pro, payload:  Geometrics MagArrow on 3m long suspension cords, flight altitude 4.5m. Planned distance between sensor and water 1.5m. Pretty calm.

The orange line denotes the altitude from the radar altimeter, whereas the blue line denotes the altitude from the barometric altimeter (in reality fused altitude using data from IMU).

Not bad. There is some drift of barometric altitude, but on average it remains the same. Pretty safe.

Spikes in radar altimeter measurements were due to the blocking of the altimeter’s beam by a swinging sensor. We were in the middle of the process of terrain-following system tuning for MagArrow.

#2. The same day, same drone, same payload, same place, same altitude. Last flight. The temperature was getting colder quickly after the sunset and the wind started to blow. Barometric altimeter drift became dangerous as it exceeded the distance between the sensor and water surface.

#3. Another day. M600 Pro, payload – echo sounder on the rope, flight altitude 2.5m.

Altimeter drift is around 2m, but in this case in the “safe” direction, as it reports lower values.

#4. And the last one, an absolutely extreme sample. M210 V2 RTK, payload: echo sounder, as before.

Barometric altitude drifts by more than 17m during a 12min flight.