DJI Mavic, Air and Mini Drones
Friendly, Helpful & Knowledgeable Community
Join Us Now

Ground to bird height not accurate

Thats why when hand catching you grab it from underneath with a thumb and fingers and dont let it autoland/descend on its own. You grab it securly then hold left stick to kill the props *while its hovering safely just above head and face height*
This is why I don't want to hand catch. Link to pictures of hand catching incidents. In the aviation world, there is a saying. There are old pilots and bold pilots, but there are no old bold pilots. FWIW.
 
Its not GPS - its the barometer and when low enough. The drone doesnt use GPS for altitude.
And its a cheap barometer that doesnt appear to be temperature compensated. It'll drift. Its quite common to see -15ft on landing for example.

If youre referring to the VPS that cuts in when low, again its not a precision sensor. Don't worry about it.

Exactly
 
  • Like
Reactions: Thomas B
The difference of a half meter isn’t that important so why worry about it?
Why are you worried that I'm worried? Maybe 20 ft off isn't that important to you. If so, don't worry about people like me that worry about it. Carry on. Some people on this forum are interested in things like racing stripes for their drone. To each his/her own.
I'll tell you why I worry about it. Let's say you flew over a building at 10 ft above it. Your next pass while maintaining the same 'indicated' altitude might not clear the building? Perhaps instead of -20 ft off the barometer is +20 ft off! o_O I know, use your camera for clearance or the collision sensors will save you (unless you happen to be flying sideways in Position mode). For all the Alfred E Neumans out there, don't worry, be happy and carry on. I guess DJI is 'worried' about it as they are updating the firmware to address this 'worrisome' problem. The drone business is competitive so this should be in DJI's interest not to mention that accuracy is always a desirable goal especially in expensive drones wouldn't you say?

From Drew at DJI tech support:
"Regarding the altitude, information is mainly obtained from the barometer. As the barometer’s accuracy is not that great, and the altitude measurement is influenced by the environment, although measurement error will occur.

It is anticipated that the altitude information will be added in the next firmware.

For the meantime, we normally provide troubleshooting for the IMU as it fixed the issue with the barometer. If not, as what is mentioned earlier it will be added on the next firmware update." :)
 
Last edited:
Well it's not that important to me but it would be nice if DJI integrated gps altitude into the barometric for improved accuracy. Maybe the gps altitude info is not as accurate as baro? It is 'detuned' for civilian use. Farmers use a gps system that is very accurate but requires multiple receivers that compare to each other. I'm not sure what impact this error has on RTH as I haven't tried it yet. Maybe I will later today? It seems to be enough of a problem that DJI is putting out an update to help with this issue. Warming up the engine before flying sounds like a good idea. I'll try it. I think the problem will still be there though as I have notice it even with succeeding flights with little delay between battery swaps.
Vertical GPS altitude info is notoriously inaccurate because of the nature of GPS satellite use. Barometric pressure is far more accurate, but requires an accurate reference point, and stable air.
 
  • Like
Reactions: Rangerider
Vertical GPS altitude info is notoriously inaccurate because of the nature of GPS satellite use. Barometric pressure is far more accurate, but requires an accurate reference point, and stable air.

Just to clarify, GPS altitude AMSL is far more accurate than barometric, due to variation in the density of the atmosphere from the standard model. When it comes to accuracy of relative height for the duration and range of a typical flight the picture is more complicated.

GPS vertical absolute accuracy is typically 1.5x worse than than horizontal accuracy. With WAAS that means it is almost always within 10 meters random error, but may be worse without WAAS. GPS relative accuracy is better but varies as the constellation movement superimposes geometric errors.

Barometric relative altitude accuracy is limited by random errors and the resolution of the sensor, together with systematic errors from the deviation of the atmosphere from standard. The systematic errors may be fairly constant over the duration of a flight, or can change by as much as 20 meters equivalent in some weather conditions.

The bottom line is that I've looked at flights where the GPS altitude data were better than the barometric, and vice versa.
 
  • Like
Reactions: Concern
Its not detuned for civilian use. Its a problem with the earth not being a perfect sphere.
I sit corrected. thanks. They (the gov) used to 'detune' the signal in the 1990's but now they don't. Real time accuracy of a few centimeters is possible with dual gps receivers! This is probably how farmers are getting such accuracy in their fields, they use multiple receivers.

I'm just guessing on this, but might it be possible for DJI to use receivers in both the drone and the controller thus utilizing dual receiver accuracy?

I found this link on selecting GPS for sUAV's quite interesting.
 
Last edited:
but might it be possible for DJI to use receivers in both the drone and the controller thus utilizing dual receiver accuracy?
It's not exactly like you are wishing.

DJI has aircraft for professional use which are equipped with RTK hardware. Those RTK systems are able to provide accuracy horizontally of ~1 cm and vertically of ~2 cm.

These aircraft are the Phantom 4 RTK and Matrice 210 RTK as well as an add-on RTK system to be mounted on any other heavy-lift drone.
 
Several quick thoughts:

1. If the drone is 100 feet away, you may think the ground is level between takeoff point and that place 100 feet from you, but unless you have a surveyor's transit and laser level, you can't be sure. Minor slopes are tough to detect, and you might very well have a difference in altitude of several feet. There are all sorts of optical illusions which can make you think two points are at the same height, when they are not. I've seen this with my own eyes at a local tourist attraction called the Santa Cruz Mystery Spot.

2. The "de-tuning" mentioned is called "selective availability" and was constantly done by the USA government so our GPS system could not be used against us. Errors were fed into the system so the bad guys' missiles, using our own GPS against us, would miss the mark. The errors in altitude were extremely severe, often amounting to 100-200 feet, or more.

Even after selective availability was ended, towards the end of the Clinton years, the altitude accuracy was never that great, and it still drifts some. You might very well see this if you measure altitude over a known point at the beginning of your flights, and then again at the end of a fifteen minute or longer mission.

Finally, one person mentioned "dual GPS." I don't know about that, but if you want accuracy down to a centimeter, you invest in a differential GPS receiver. I don't know if this can be hacked into any DJI drones, but I have an old Garmin that can use it. The way it works is you feed your GPS a correction signal from a nearby GPS receiver that is sitting directly on a reference monument or benchmark (i.e., a location whose lat/long/altitude are perfectly known). If the reference point is reasonably close (within a few miles) all the little errors in the GPS signal from satellite perturbations, propagation delay, temperature changes, will affect both points in the same manner. Because of these imperfections, the reference GPS will see itself moving around, when it obviously knows it is not. It uses this phantom motion to send out a correction signal to your GPS receiver.

Using this technology, you can get GPS measurements that are accurate to within centimeters, or less.

Read about it here:

Differential GPS
 
Several quick thoughts:

1. If the drone is 100 feet away, you may think the ground is level between takeoff point and that place 100 feet from you, but unless you have a surveyor's transit and laser level, you can't be sure. Minor slopes are tough to detect, and you might very well have a difference in altitude of several feet. There are all sorts of optical illusions which can make you think two points are at the same height, when they are not. I've seen this with my own eyes at a local tourist attraction called the Santa Cruz Mystery Spot.

2. The "de-tuning" mentioned is called "selective availability" and was constantly done by the USA government so our GPS system could not be used against us. Errors were fed into the system so the bad guys' missiles, using our own GPS against us, would miss the mark. The errors in altitude were extremely severe, often amounting to 100-200 feet, or more.

Even after selective availability was ended, towards the end of the Clinton years, the altitude accuracy was never that great, and it still drifts some. You might very well see this if you measure altitude over a known point at the beginning of your flights, and then again at the end of a fifteen minute or longer mission.

Finally, one person mentioned "dual GPS." I don't know about that, but if you want accuracy down to a centimeter, you invest in a differential GPS receiver. I don't know if this can be hacked into any DJI drones, but I have an old Garmin that can use it. The way it works is you feed your GPS a correction signal from a nearby GPS receiver that is sitting directly on a reference monument or benchmark (i.e., a location whose lat/long/altitude are perfectly known). If the reference point is reasonably close (within a few miles) all the little errors in the GPS signal from satellite perturbations, propagation delay, temperature changes, will affect both points in the same manner. Because of these imperfections, the reference GPS will see itself moving around, when it obviously knows it is not. It uses this phantom motion to send out a correction signal to your GPS receiver.

Using this technology, you can get GPS measurements that are accurate to within centimeters, or less.

Read about it here:

Differential GPS
Thanks that is very interesting. That brings up my proposal/question if 2 gps units could 'correct' the errors of the barometer and gps altitude measured by the drone unit gps. If another gps unit was located inside the smart controller, the position/altitude would be pretty much static in terms of altitude thus a known reference point for corrections which could be sent to the drone gps/imu calculated position to correct the altitude only. Wouldn't that do it? Hey dji, put another gps in your smart controllers and allow the PIC to input altitude or lock the initial altitude into the controller as read by the drone to 'zero' the altitude in the controller as the correction input to the drone.
It is not uncommon for me to be flying on varying elevation terrain of +- 100 ft or more.
 
Last edited:
Thanks that is very interesting. That brings up my proposal/question if 2 gps units could 'correct' the errors of the barometer and gps altitude measured by the drone unit gps.
If another gps unit was located inside the smart controller, the position/altitude would be pretty much static in terms of altitude thus a known reference point for corrections which could be sent to the drone gps/imu calculated position to correct the altitude only. Wouldn't that do it?
No that wouldn't do it.
Your idea about a second GPS is differential GPS but there's a lot more to differential GPS that just using two GPS units, look it up to find out how it works.

Hey dji, put another gps in your smart controllers and allow the PIC to input altitude or lock the initial altitude into the controller as read by the drone to 'zero' the altitude in the controller as the correction input to the drone.
All flyers except some using iPads already have a gps receiver in their controller/phone/tablet.
 
  • Like
Reactions: Rangerider
Okay it looks like I need to trade up to an RTK Vbox drone! :) Or I can just wait until the following happens?

"'L2C' – Unencrypted ‘civilian’ signal on the L2 freqency, which is more powerful than L1. This has been added to satellites launched since 2005, and is currently present on around 19 satellites. While this has not yet reached a full constellation of 24 satellites, the frequency is transmitting and can be used. When a full 24 satellites are available, this will bring down the accuracy of GPS positional data (without the use of a base station) from ±3 m to ±1 m. This frequency should be fully functional within the next couple of years. "
 
Last edited:
Thats why when hand catching you grab it from underneath with a thumb and fingers and dont let it autoland/descend on its own. You grab it securly then hold left stick to kill the props *while its hovering safely just above head and face height*

You won’t ever catch me reaching into a ****** off Salad Shooter... I’ve seen too many people slip up and darned near lose a finger or worse....
 
Thats why when hand catching you grab it from underneath with a thumb and fingers and dont let it autoland/descend on its own. You grab it securly then hold left stick to kill the props *while its hovering safely just above head and face height*
This is a bad idea for two reasons. First, it won't be long before you get nailed by a blade. Second, there is no good reason to stress your motors like that. Unless your hand is just tiny, it is far better to allow the aircraft to 'see' your palm as a solid landing spot and simply land it on your flat, outstretched palm as you normally would, with left down stick in Mode 2. There are YT vids that show this method. I have been using it for over a year without mishap, although occasionally the bird will drift slightly and not 'see' your hand, it gets easier with practice. No more fingers in the blades, and the UA shuts down normally, without redlining the motors. I fly a MP1 and it gets a little warm on the bottom of the bird after a max-duration flight, but not enough to burn my palm. Once you try it, you'll never go back! Cheers.
 
  • Like
Reactions: Rangerider
The motors undergo no stress at all. You're not holding it, not precenting it moving, you're gripping it without moving until you kill the motors with the other hand. Its a hell of a lot safer than a spinning drone descending automatically towards your hand which might be moving (on a boat etc) and a drone which is wobbling in the air as it comes down.
Theres no way in **** im going to use the landing on a palm method on land and certainly not on a boat which might be rolling slightly or a gusty windy day. I dislike the whole concept of flying (or worse it enters auto landing) towards a body part because you have no direct control at all over it at that point.
I want the thing with spinning props staying exactly where it is and i move around it, not the other way. Theres far less chance getting nailed by a blade by you going to a stationary drone rather than a drone flying/descending towards you.
 
  • Like
Reactions: Mike4884
Second, there is no good reason to stress your motors like that. ... the UA shuts down normally, without redlining the motors.
I'm curious about why you'd think that hand catching stresses the motors?
And if you think catching is bad for the motors, wouldn't flying be even worse?
 
  • Like
Reactions: Mike4884
This is a bad idea for two reasons. First, it won't be long before you get nailed by a blade. Second, there is no good reason to stress your motors like that. Unless your hand is just tiny, it is far better to allow the aircraft to 'see' your palm as a solid landing spot and simply land it on your flat, outstretched palm as you normally would, with left down stick in Mode 2. There are YT vids that show this method. I have been using it for over a year without mishap, although occasionally the bird will drift slightly and not 'see' your hand, it gets easier with practice. No more fingers in the blades, and the UA shuts down normally, without redlining the motors. I fly a MP1 and it gets a little warm on the bottom of the bird after a max-duration flight, but not enough to burn my palm. Once you try it, you'll never go back! Cheers.
Why would the motors undergo stress this makes no sense only thing I could think of is if you didn't turn off landing protection in vps settings yes the aircraft will try to fight you but if you turn this off it will turn off super easily.... Also they are brushless motors they won't be stressed out from just you turning off your quadcopter also I have never hit one of my fingers hand catching.... And guess what my quadcopter still looks brand new instead of landing in mud, the hard Street or whatever else. Maybe man up and try one or two times you wont go back fella.it is pretty simple to do just don't put your fingers where the blades go!!! Also I have seen 2 people try to catch it with their Palm flat.... yeah not a good idea it fell over on them. the one guy it looked like there was a gust of wind so the motors try to compensate and kind of revved up a little bit while drone was on his palm, and this is why it fell over. so I prefer to grab mine from the bottom. I'm sorry dude but that is really horrible advice you are giving.
 
I hand catch like in the video below. Basically, you set it to hover where you can reach it and grasp it in front of the downward sensors while it maintains a hover. It does not detect your hand so it will not try to rise up so you do not need to turn off the downward sensors. Then while holding it push left stick down. The MP does not detect more downward movement so it assumes it has landed and shuts off the motors.

Since my MP is so stable in a hover, I feel more comfortable reaching up above my head and grasping it than letting it settle on my hand. To each their own, but this works for me.


EDIT: Sorry about this OP. I did not mean to hijack your "height not accurate" thread. Hand catching opinions may be found by searching the forum.
 
Last edited:
Lycus Tech Mavic Air 3 Case

DJI Drone Deals

New Threads

Forum statistics

Threads
130,586
Messages
1,554,118
Members
159,586
Latest member
DoubleBarS