Large discrepancy between VPS and baro altitude

[AyeYo]

Had a job last weekend where a client wanted shots from 30ft AGL to determine views from a possible second story addition. This should have been a piece of cake, but it wasn't.

I took it up, flew to level with the top of his current roof, over clear ground and looked at the telemetry. The height showed 25ft, so I flew up to 30ft. 30ft looked really low, so I looked over VPS and saw only 18ft. I flew to 30ft on VPS and was at over 68ft according to the altitude reading. This continued, all over the yard as I tried to get 30ft shots, never knowing which reading to believe and finally just going off a visual guess of where a second floor would be.

Is this normal? I never needed that kind of altitude precision before so I simply ignored VPS height and always looked at the altitude reading. But the more situations I've looked at the altitude reading it, the more I check altitude while flying around the more obvious it's become that it's not even remotely accurate.

 

[msinger]

Between the two, the VPS altitude is usually more accurate since it's determined using sensors that can see/detect the ground (or obstacles) beneath the Phantom. The altitude from the IMU is estimated using data from the barometer. It's also the altitude above the takeoff location, so a change in elevation can affect it if you're not flying directly over the takeoff location.

 

[AyeYo]

Between the two, the VPS altitude is usually more accurate since it's determined using sensors that can see/detect the ground (or obstacles) beneath the Phantom. The altitude from the IMU is estimated using data from the barometer. It's also the altitude above the takeoff location, so a change in elevation can affect it if you're not flying directly over the takeoff location.

What's the range of the downward VPS? It sees to stop registering around 40ft.

 

[msinger]

Per the Phantom manual, it's effective between 0.3 to 10 meters.

 

[Richard R]

Page 6 of the manual that I have (v1.6) says that VPS is useable from 0.3 to 3 meters (10 ft). The bar should be pretty accurate over the times that you can fly on a battery. But like msinger said, its referenced to the take off location. Simple enough to eyeball any change in ground elevation between the takeoff position and the the ground around the house. I'd use the reported altitude not the VPS for anything over 10 ft.

Oops, Didn't pay close enough to where this was posted, I was reading from the P3 manual. Dont know if the VPS is different on the P4, but I would still go with the baro reading.

 

[msinger]

Oops, Didn't pay close enough to where this was posted, I was reading from the P3 manual. Dont know if the VPS is different on the P4, but I would still go with the bro reading.

Yes, it's different for the P4P.

 

[Richard R]

Just read the P4 manual - Msinger was correct, the P4 VPS is supposed to work up to 10 m, but that would be under ideal conditions so by the time you get to 25-30 high, you are right on the fringe of accurate measurements and discrepancies between VPS and Baro would be expected.

 

[DanMan32]

Barometer can be off a few feet over the course of time the battery would allow.
I watched video recordings for a 20 minute run where I have close caption enabled. Barometer is set to zero as soon as the motors are shut off. It was at -1.5 on touchdown on two flights, where I estimate a reading of 30 is about 100ft. That leaves a difference of roughly 5 feet over the course of 20 minutes. It was late afternoon so cooling temperatures within a mile of the bay shore could account for the barometric change.

 

[AyeYo]

Barometer can be off a few feet over the course of time the battery would allow.
I watched video recordings for a 20 minute run where I have close caption enabled. Barometer is set to zero as soon as the motors are shut off. It was at -1.5 on touchdown on two flights, where I estimate a reading of 30 is about 100ft. That leaves a difference of roughly 5 feet over the course of 20 minutes. It was late afternoon so cooling temperatures within a mile of the bay shore could account for the barometric change.

That doesn't really explain a 100% margin of error over the VPS that I experienced. I'll have to pay closer attention to it at landing time.

 

[N017RW]

Sure it does. The baro is likely very inexpensive, poorly filtered, and subject to ambient conditions.
Then you're comparing it to differing technology (sonar).
If you had two identical digital watches you'd begin to see drift between them in 24-48 hrs and continuing as time elapses.

Thus Confucius say: 'Man with one watch always knows what time it is, Man with two watches never knows what time it is.

 

[SkyHog]

That doesn't really explain a 100% margin of error over the VPS that I experienced. I'll have to pay closer attention to it at landing time.

Yes, the Baro drifts a little, as N017RW said above. But you may also be underestimating the altitude difference between takeoff point and the area you are flying. Remember, the barometer is not measuring height above the ground below. It is measuring height above takeoff point. The terrain in your yard may vary a lot more than it appears. Most yards are not perfectly flat. Mine varies ten to 20 feet easily between one area and another within a 2 acre lot, even though it appears to be generally flat. And the house is at the highest point with gentle slopes away from the house, which would explain your observations.

 

[AyeYo]

Yes, the Baro drifts a little, as N017RW said above. But you may also be underestimating the altitude difference between takeoff point and the area you are flying. Remember, the barometer is not measuring height above the ground below. It is measuring height above takeoff point. The terrain in your yard may vary a lot more than it appears. Most yards are not perfectly flat. Mine varies ten to 20 feet easily between one area and another within a 2 acre lot, even though it appears to be generally flat. And the house is at the highest point with gentle slopes away from the house, which would explain your observations.

The yard was dead flat. I flew straight up from takeoff spot. 30ft on VPS showed 68ft on baro. That's an insane margin of error within 10 seconds of takeoff.

 

[N017RW]

Yea. But it's just a Chinese toy.

 

[DanMan32]

I guess I wasn't paying attention. 30ft and 68ft for same actual altitude is quite a discrepancy, especially if taken within a short time after takeoff.
VPS might have been deceived by a stray sonar echo. P4 has a higher operational range than a P3 but it's accuracy may diminish at ranges such as 30ft.
About the only way to know for sure is check with an object of already known height or determined by trigonometry.

 

[tcope]

VPS stops working at 32 feet. If the barameter was correct and you were at 68', the VPS may have been simply giving you an incorrect "ghost" reading. It may not stop giving a reading at close to it's limit.

My recommendation would be to calibrate the IMUs at the flight location for the best reading.

 

[SkyHog]

A good way to test is to launch vertically over the takeoff point, and watch/record both sensors. It isn't unusual to see a few feet variance in the baro sensor over a flight time, due to actual changes in atmospheric pressure. Even prop wash and flight speed could affect a sensor like that due to Bernoulli effects, if you are actually trying to measure heights with single digit accuracy. However, the Phantom baro sensor is actually quite good for its intended purposes, which generally don't require single digit accuracy for relative height measurements, much less absolute height measurements. The VPS sensor should be even more repeatable and accurate, but only within a narrow range close to the ground.

A good way to characterize inherent short term relative accuracy of these two sensors is to just hover over the takeoff point and record what you get as altitude increases, until the VPS gives up and doesn't correlate well anymore.

 

[AyeYo]

I guess I wasn't paying attention. 30ft and 68ft for same actual altitude is quite a discrepancy, especially if taken within a short time after takeoff.
VPS might have been deceived by a stray sonar echo. P4 has a higher operational range than a P3 but it's accuracy may diminish at ranges such as 30ft.
About the only way to know for sure is check with an object of already known height or determined by trigonometry.

Judging by the height of the house, it was the VPS that was accurate at 30ft and the baro that was lightyears off. I'm going to pay better attention to them on my next flight and see how they do (or don't) correlate.

 

[DanMan32]

I just figured out. The barometer reading in the subtitle file is height in meters. Doh!

 

[With The Birds]

Between the two, the VPS altitude is usually more accurate since it's determined using sensors that can see/detect the ground (or obstacles) beneath the Phantom. The altitude from the IMU is estimated using data from the barometer. It's also the altitude above the takeoff location, so a change in elevation can affect it if you're not flying directly over the takeoff location.

Does the IMU on the P4P contain the barometer? The only IMU's I have seen have gyro and accelerometer (on rare occasion a magnetometer). Might've another clue how DJI are keeping cost relatively low while buying in the latest tech.

 

[N017RW]

Loosely speaking yes, the IMU contains the Baro.
In reality it is not co-located physically.