The user interface of the M300 shows the wind speed and direction, which is based on feedback from the motor.
Can I get these values from the MobileSDK?
The values for the simulator can be set, but I haven't been able to find the corresponding API.

It would be great to get it in Mavic2 as well.

I found WIND_SPEED key and WIND_DERICTION key in API document. But it doesn't seem to work well on M300.
Link: https://doesn'tdeveloper.dji.com/api-reference/android-api/Components/KeyManager/DJIFlightControllerKey.html?search=wind&i=0&#flightcontrollerkey_wind_speed_key

kv886 Posted at 6-15 04:27
I found WIND_SPEED key and WIND_DERICTION key in API document. But it doesn't seem to work well on M300.
Link: https://doesn'tdeveloper.dji.com/api-reference/android-api/Components/KeyManager/DJIFlightControllerKey.html?search=wind&i=0&#flightcontrollerkey_wind_speed_key

I found it, too, but the "Access" was a "PUSH", so I'm sure that's what sets the value to the simulator.

z37soft Posted at 6-16 02:48
I found it, too, but the "Access" was a "PUSH", so I'm sure that's what sets the value to the simulator.

Should I email such a request to dev@dji.com?

Sorry for necroing, but did anything ever come of this discussion/request?
I work for a company using DJI Drones as well, and would LOVE for this feature to be implemented!

- Andreas

No, there's no progress.
I believe it can be calculated by software based on no load and the effect of wind pressure.
It would be nice if DJI would allow us to access the internal information through the SDK...

z37soft Posted at 6-7 04:29
No, there's no progress.
I believe it can be calculated by software based on no load and the effect of wind pressure.
It would be nice if DJI would allow us to access the internal information through the SDK...

I've done some wind meters.

You have two option.
1 Use the attitude of the aircraft. Calc windspeed from gps-speed.
2 Let it drift in the wind, flat level and turn yaw slowly. Wait for the speed to stabilize and you got a mean value.

2 Is the most accurate, but can't be done under normal flight.

1 Needs to be adjusted for specific drone type and/or payload / air temp / altitude, so it's much harder, but it would be nice to have. You could do som calibration functions, like checking speed for different attitude. Please write here if you do that. I'm interested in the data.

Airdata does 1, but i found it not very accurate.

The wind keys you talking about is for the sim. I never managed to set them.

Thanks.
I thought the "attitude" value was the value after resisting wind and disturbance. If we can read from the "attitude" that it is resisting the wind, then this method will work.

The M300 does the same thing, so I'm sure it can be done with other products as well. Because we believe that the values are obtained from a part of the drone control without using any special sensors.

z37soft Posted at 6-7 17:17
Thanks.
I thought the "attitude" value was the value after resisting wind and disturbance. If we can read from the "attitude" that it is resisting the wind, then this method will work.

Bear in mind you have to calc some mean value from attitude. (And roll)
Sampling rate seems to be 10Hz on the mavic2
It can vary quite a lot, depending how gusty and and turbulent it is.

I actually though of present a figure of how gusty/turbulent the wind is, it's quite important for kitesurfing which is my usecase :-)

I do most of my flights in very high wind, and it's not uncommon that the attitude maxes out. So that might be a special case, I don't know really.

I use it for windsurfing. I wanted to know how strong the wind is offshore, a couple of kilometers from shore, rather than on the beach. It's the same situation.

z37soft Posted at 6-9 16:08
I use it for windsurfing. I wanted to know how strong the wind is offshore, a couple of kilometers from shore, rather than on the beach. It's the same situation.

Cool, exactly my usecase. I never pick the wrong kite any longer :-)

In that case I recommend drifting with the wind and slowly rotate at the same time. Rotation is due to that tha attitude is never completly correct.

Have you seen my followme app for surfing? It waits until I'm 800m out and starts to followme.

Awesome. Not really the topic here, but some interesting stuff.
First, the subject is in the center in most cases. That's good.
What surprised me the most was that the drone was in front of person. How is this being tracked?

z37soft Posted at 6-10 06:52
Awesome. Not really the topic here, but some interesting stuff.
First, the subject is in the center in most cases. That's good.
What surprised me the most was that the drone was in front of person. How is this being tracked?

I wear a 2:nd phone on my helmet that sends gps data to the phone which is connected to the remote on the beach.

I tag the gps data with the gps timestamp, and use that to calulate the the delay time over 4G fo reach package. It then uses dead reckoning to predict the path of the tracker.

It also uses the free gimbal mode hack to get smooth gimbal movements, which is crucial for filming in high winds. (in virtual stick mode)

The gps data is filtered but not filtered at all for big movements. It uses the android built in GPS fuse mode together with dual freq gps in the phone.

Uses zerotier which uses punch through teqnuiqe to get the latency down in cell network. Sends just enough data to optimize the latency. Latency is only between 40-100ms. UDP. Direct peer to peer.

Everything made in less than 300lines of python :-) Python bindings framework not included.

Next step is to go lower and closer and using opencv from the live view and to adjust for gps-errors.

You have taken a similar approach.
I'll post a use case for windsurfing here on "Share" later.

There was a time when I put a cell phone in the back of a player in the same way.


I think the last position adjustment is still limited by GNSS and needs to be adjusted by image processing.
Here's what it looks like.


From the latency, I got the impression that your method is smarter in the process of looking ahead.
I'll try to improve my algorithm some more!

z37soft Posted at 6-10 17:14
You have taken a similar approach.
I'll post a use case for windsurfing here on "Share" later.

Super interesting videos!
We seems to be doing very similar things :-)

Autopilot drone for wide-area sports


（I'm using VirtualStick Mode, which the developers love.）

We have developed autopilot software by utilizing the sub-meter class positioning augmentation
service, one of the services provided by the Quasi-Zenith Satellite System, to utilize position data that is more accurate than GPS.

Drone：DJI Mavic2Zoom
Our software：Android / DJI Mobile SDK
Location service：QZSS L1S SLAS

With the cooperation of N-Sports tracking Lab LLC, each athlete was equipped with a SLAS-compatible receiver, and the information was used to realize an automatic shooting drone.

At sea, where there are no landmarks, the pilot, the drone operator, must fly by watching FPV.
In this situation, it is very difficult to keep track of the ever-changing race conditions in real time and to know who is running where.

In this project, we extracted "highlights" such as points where the distance between athletes is close and where they are competing, based on their location information.
The drone was placed in the "highlight" area by automatic flight, and automatic photography was attempted.

As a result, we were able to accurately confirm the location of the players and send the drone to that location.
We have also confirmed that the drone can automatically determine "highlights" and automatically shoot at points that are likely to be highlights.
In the future, we hope to further improve the quality of photography and reduce the burden on the drone operator in harsh environments.


About QZSS
The Quasi-Zenith Satellite System (QZSS) is a regional navigation satellite system that
transmits additional GPS L1 C/A signals for the Pacific region covering Japan and Australia.

About QZSS L1S SLAS
QZSS SLAS (sub-meter level augmentation service) is an augmentation technology which provides
correction data for pseudoranges of GPS and QZSS satellites.

Thanks.

z37soft Posted at 6-11 23:19
Autopilot drone for wide-area sports

https://youtu.be/orQYDYzNmIU

Fantastic, didn't know this.

Can you explain how the tracker hardware is, and how is it transfered to the remote?
Do you use many trackers on each sailor, or just one?

Is the app released yet?

Thanks!

For the hardware, we developed a device with u-blox, a LTE module, and a battery inside.
The communication is LTE. For small areas, UDP is much better.
There is one for each sailor. We have many of them available.
There are no plans to release applications at this time. We are in the process of repeating PoC.

z37soft Posted at 6-12 06:09
For the hardware, we developed a device with u-blox, a LTE module, and a battery inside.
The communication is LTE. For small areas, UDP is much better.
There is one for each sailor. We have many of them available.

Thanks!

You are using LTE as well :-)