Re: [Paparazzi-devel] Barometer/Altitude Estimate Drift

[Luke Ionno]

One clarification, using the “no_type” INS, exactly what sensor readings are fused into the altitude estimate?  Is the GPS used only to zero out the barometers during initialization, or does it remain an active part of the filter input throughout operation? 

 

-Luke

 

From: address@hidden [mailto:address@hidden On Behalf Of Luke Ionno
Sent: Saturday, March 08, 2014 9:36 AM
To: 'Paparazzi UAV devel list'
Subject: Re: [Paparazzi-devel] Barometer/Altitude Estimate Drift

 

Thanks for the information!  I’ve actually got a pair of LISA/M 2.0s, and having done some multi-ship work with them, I know they both experience the downward drift, though come to think of it, one of them may have a bit less drift than the other.

 

I just did a flight test with a 1 hr. ‘preheat’, and it seemed to help, but in the end, I still had ~3m downward drift in the estimate over an 18 minute flight.  During the preheat, the altitude estimate was varying by 2+ meters, so if the motor noise is contributing, it’s not changing things that much.  (I’ve already got a sunshade on the sensor.)

 

In the end, I think Kadir is right, sonar is the way to go for my application.  When I’m indoors, I’m flying over a hard, flat floor, which should provide a good strong return on the sonar pings...  Outdoors, I’ll just leave 3-4m of altitude margin, and accept the drift.

 

-Luke

 

From: address@hidden [mailto:address@hidden] On Behalf Of Prof. Dr.-Ing. Heinrich Warmers
Sent: Saturday, March 08, 2014 7:20 AM
To: Paparazzi UAV devel list
Subject: Re: [Paparazzi-devel] Barometer/Altitude Estimate Drift

 

Hi Iomo,
we had also  a drift problem with this sensor.
After we change the sensor we had no Problems. So the sensor was defect.
With the Bosch BPM085 i had many problems after soldering and head up.
About  30% of the sensor i had to change.
For pressure sensors on muticopters  you can apply  some foam  if you have problems with the Propellor wind.
This also help for light effects.
Regards

Heinrich
 

Luke Ionno schrieb:

Hello all,

 

I need a reality check on exactly how good (or bad) an altitude estimate I should be expecting from the V5.0.3 rotary wing branch, using the ‘no_type’ INS, and a LISA/M 2.0, with the MS5611 barometer on an Aspirin 2.2 IMU board.

 

Right now, in a completely static indoor environment (no fans, open windows, etc.)  I’m seeing upwards of 2 meters of altitude estimate drift (always upwards, I might add).  The attached files Static Test 1.jpg and Static Test 2.jpg show roughly 20 minutes of static (motors off) testing; I reset the vertical filter after the end of Static Test 1.jpg, but didn’t cycle power.  Every minute or so, I’d lift the quad up to the ceiling (~2.25 meters up), hold it there for ~10 seconds, and then set it back down.  Over the first 10 minutes, there’s approximately 2.5m of upwards altitude drift, and then another 1.75m over the next 10 minutes.  The overall magnitude of the floor-to-ceiling step input seems consistent, but the absolute altitude estimate drifts quite a bit. 

 

I then took 10 minutes of raw barometer readings, which showed a similar trend, as shown in Static Raw Baro.jpg.  At the end, I threw in a couple floor-ceiling cycles, just to get an estimate of how much drift I was seeing.  (If I were using the raw barometer readings to estimate altitude, the drift over 10 minutes would have been ~1 meter.)  Of course, that’s all with motors off; I did an outdoor tethered-hover test, with a 2m tether, in manual flight mode, recording the raw barometer readings, and got the plot shown in Tethered_Hover_Raw_Baro.jpg; note that at each landing event, the barometer reading has drifted by ~2m in a matter of 60 seconds.  The barometer is well clear of the ESCs and other potential heat sources, and I allow it to acclimate to the outside temperature before flying. 

 

I’ve flown a number of NAV flights with this setup, and they’re quite consistent with the static plots and tethered tests; I’ll start out say, 2-3m AGL, and over a minute or so, it’ll drift downwards by a  couple meters, until it ends up sitting on the ground with the motors idling.  (The vertical loops are well-tuned, it’s that the altitude estimate itself develops errors).  I can bump it up a couple meters, and then it’ll hold for somewhat longer, but it general trends downwards over the course of a 15-20 minute flight by ~4m or so.  As noted before, the aircraft altitude estimate always up, never down.

 

So, are the various drifts shown above normal for a GPS + barometer setup?  Is a +/- 2m altitude estimate simply too much to ask of the system, or do I need to start hunting down sources of vibration/noise, etc.?  Is anybody else experiencing this sort of behavior?

 

Thanks!

 

Regards,

-Luke

 

 

 

 

 

 

 

 

---

 

 

---

 

 

---

 

 

---

 

_______________________________________________

Paparazzi-devel mailing list

address@hidden

https://lists.nongnu.org/mailman/listinfo/paparazzi-devel