Was running pix4d, mission completed, GO_HOME signal sent, drone took 40 seconds to get to homepoint, then 90 seconds to land from 45m. Drone wasn't found at homepoint and we are trying to figure out where it might be. No obstructions 70ft from homepoint or on path to homepoint from last known location.

We suspect the tablet mounted to the controller flopped down onto the controls. This would have pinned the two controlls up during most of the mission and while the drone was in the RTH state.

In a case where RTH is issued I suspect the behavior would have been as follows:1) adjust alt up if required (it should have been above the 30m set homepoint alt given the mission was 45m)
2) Point to homepoint
3) Fly straight line to homepoint (At this point I believe throttle is the only control which affects drone?)
4) Stop at home point and decend (also a point where I believe controls have an effect on drone ... all controlls in this case? )

Further Question, If the drone is at homepoint and gets a alt up command does it climb or resist climbing?



Assuming the above question is YES then would it be fair to assume the drone flew both UP and Forward while in LANDING mode (via dji-ctrl logs) before reaching a state of LANDED ... which would likely mean it ran out of battery and decided to do an emergency landing from way high up in the air ...

This might help us find the drone as the direction forward of the drone for miles is forest ... so while it wouldn't be easy it might be possible to recover the drone if it flew there.

The only other thing we can think of right now is the drone landed >70ft from the home point ignoring the pinned sticks (which were likely pinned the whole mission) and landed in a location where it was burried with dirt we were having delivered that day. (I haven't dug it up yet to see if it is there or not)

I realize pix4d is not a DJI supported application so only seeking advice on RTH behavior, I have posted in the Pix4D forums as well but only questions in relation to the application and suggested features to log location of the drone at all times till it is landed.

The DJI find my drone feature has the drone right at the homepoint when I tried to use it to find the drone.

Before answering your questions, I would just like to confirm what I understand. Basically, you took off the drone, put down the remote control and left to do something else. And when you came back the drone wasn't where it should be?
I don't know the PIX4D flight plan application at all. In principle, a phantom on a mission, no matter what commands you put on Throttle, pitch, roll, the drone will do nothing. If you put yaw he will execute. Then in the landing phase you can act on pitch roll and yaw in order to refine the place of landing.
You can also act on throttle down to accelerate the descent. I have never tried to throttle up! but what is certain is that if pitch up is active, the drone will move.
If you know the path of the drone, draw a line on a map and start looking along that line, that's what I'll do. Finally no, I will start by taking care of my drone during the flight, but that is another story.

I just tried, yes indeed by folding the tablet the two sticks are pushed forward.
90 seconds to descend 45m does not seem abnormal for an automatic landing, the drone must not be far away!
If you remember the orientation of the drone at departure, look in that direction. Because in the test I did, the drone started by taking this direction at end of RTH, before starting the descent.
I'm betting about 400m in that direction if he didn't hit an object first. The anti-collision system is not friends with trees!
good luck !

I was plotting the straight line in my QGIS application when I got your response and was already planning to navigate that way tomorrow ... the question is how far at this point ... not sure of ALT of drone when my controls could have affected it. Trees are ~100-120ft in the forward direction of the drone when going to land ... just hope it isn't IN a tree.

The Pix4D does create a mission and executes the mission using ctrl dji api. At the end of the mission a new "return to home" mission is created but sadly this doesn't appear to have its own GPS log. (image of said mission attached, there are many other logs as well I have reviewed to help me locate the drone. my suggestion to Pix4D was to not stop capturing the GPS data till the drone is landed)

The RTH mission in my case was: (taken from ctrl logs)
SDK flying state: GO_HOME (SDK called to go to RTH)
Flying state: IN_WAYPOINT_MISSION -> GOING_TO_LAND (mission it was previously on switched to GOING_TO_LAND)
Event at 1667240479165 of type TAKING_OFF (likely a formality in the logs indicating it has taken off to start the mission, or this could be the drone getting to ALT required)
Event at 1667240497819 of type GOING_TO_MISSION (RTH starts, straight line path)
SDK flying state: AUTO_LANDING (SDK called to comence auto landing, one of three landing modes if I recall)
Flying state: GOING_TO_LAND -> LANDING
Flying state: LANDING -> LANDED
Flying state: LANDED -> UNKNOWN

When I am running mapping missions I keep the remote near me, and occasionally look in the sky for the drone to confirm location. This was the first time I didn't keep my eyes on it while it was returning from a mission.

In the API for CTRL DJI https://developer.dji.com/onboard-sdk/documentation/guides/component-guide-flight-control.html I see reference to "Users could use RC throttle’s down command(keep 2 seconds), APP's confirm landing command and SDKs confirm landing API to confirm. Conversely, RC throttle’s up command, APP cancel landing or switch flight mode will exit the landing mode(auto-landing and forced landing)." ... I am not sure if the APP has to listen for this, and if Pix4D does listen for this event, or if this is native to the drone itself.

LV_Forestry Posted at 11-3 08:07
I just tried, yes indeed by folding the tablet the two sticks are pushed forward.
90 seconds to descend 45m does not seem abnormal for an automatic landing, the drone must not be far away!
If you remember the orientation of the drone at departure, look in that direction. Because in the test I did, the drone started by taking this direction at end of RTH, before starting the descent.

My tablet has been known to "fold" on its own because the support seems weak and it can't hold it ... that day though I believe the wind might have done it for me :/

I thought 45m in 90 seconds seemed "normal" as well ... which had me thinking it is buried in the fill pile 80-150ft away vs "lost" in the woods.

Pulled out some ground lidar of the area ... drones landing point elevation is around 321m ... any elevation above or equal is red, anything from 290-320 is yellow ... since the trees are 100ft.

Going to put these points in my GPS now since it is kind of dense in there.

Just realized my map was in ft not m for contour lines so I adjusted the colors to reflect on a 8m/s flight speed only to realize the drone might have been in sport mode.

Given the logs have the drone landing from ~156ft in 90 seconds this means it is 720m to 1440m away (if we account for sport mode being active ... 8m/s vs 16m/s if I recall)  ... this also assumes the drone was only told to go forward (via the screen having collapsed onto the controller) and not being told to turn in any specific direction as well ... if the latter is true the search radius is huge.

The question would then be ... would ctrl dji's logs repot "landed" upon crashing and hitting the ground, or would it report something else if it crashed.

This thing could be up to 1 mile up the road 0__0 ... I put word out to the neighbors in hopes someone finds it. Not feeling too hopeful now.

Project Anavita Posted at 11-3 09:59
Just realized my map was in ft not m for contour lines so I adjusted the colors to reflect on a 8m/s flight speed only to realize the drone might have been in sport mode.

Given the logs have the drone landing from ~156ft in 90 seconds this means it is 720m to 1440m away (if we account for sport mode being active ... 8m/s vs 16m/s if I recall)  ... this also assumes the drone was only told to go forward (via the screen having collapsed onto the controller) and not being told to turn in any specific direction as well ... if the latter is true the search radius is huge.

I've tried to follow this complicated thread and it's not easy.
To find where the drone ended up, your best approach would be to see what the recorded flight data shows.
That should tell you exactly what happened in the flight and where the drone landed or crashed.

Airdata will accept the encoded flight data files and has instructions (for iOS or Android devices here to help you find and upload the relevant file:
https://app.airdata.com/pix4d
Upload the file to Airdata and it will give you a summary report.
Come back here and post a link to that AIrdata report and I can go into it and get the information that you need.

btw .. if the drone is is RTH, it is not in Sport Mode

Labroides Posted at 11-3 13:50
I've tried to follow this complicated thread and it's not easy.
To find where the drone ended up, your best approach would be to see what the recorded flight data shows.
That should tell you exactly what happened in the flight and where the drone landed or crashed.

Good to know the drone wont be in sport mode once it gets to the homepoint and a user takes control ... that seriously reduces the area I am going to have to search to the 8m/s range vs 16m/s range.

Sadly the FlightRecord log doesn't exist for this mission ... or any other ones I flew that day. In fact looking at the dates FlightRecord  hasn't recorded anything since 2020

The logs I have, and have reviewed, are
sync_state.json
Mission_18.log (the name of the mission)
Mission_18.gpx (can be imported into QGIS for visual)
details.pix4dcapture-mission
dataflash.log
pix4Dmapper.log
pix4dcapture.log
ctrl_dji.log
and the data you can export from the only .DAT file I saw near the mission start (which looks like it only captured my recalibration of the compass then stopped)

@Labroides

The short of it:
Drone flew a Pix4D mission and got the go home signal.
Drone flew 40 seconds back to the home point at 156ft which is above the 30m set homepoint altitude without any obstructions along the way
At this point the logs indicate the drone took 90 seconds to land and the logs indicated the drone "landed"

The issue is where did it land. My tablet was face down on the control when I when I noticed the drone was not in its homepoint on the ground. This was 10-15 minutes after it landed. This made the two controls push UP which if the DJI protocol is to allow a user to control an AUTO landing command when it is above the homepoint the drone likely speed off forward while still trying to land.

I am trying to find where it might have "landed" by calculating the time it took to land, the likely speed of the drone, the topography and environment it was crossing over, and if the logs would indicate "landed" if it "crashed" instead.

Well, you just have to follow the line you drew. Again, remember the direction the drone was pointing at the start, because at the end of the RTH it goes back to that position. So if the first line has not been successful, create another along this axis.

You couldn't be in sport mode otherwise it refuses the autonomous mission.
When the tablet is pressed against the sticks, they are barely halfway through their travel, so 8m/s not possible.

What is very strange is that you could not see the position of the drone on the tablet when it told you that it had landed. I have hit trees, when the drone hits the ground, the motors stop and it is "landed". But as long as the connection with the radio control is good, the coordinates are displayed. It is only if the battery is ejected when touching the tree that it becomes problematic. Because even if it falls into the water, you must be able to know the last point of the trajectory.

LV_Forestry Posted at 11-3 14:31
Well, you just have to follow the line you drew. Again, remember the direction the drone was pointing at the start, because at the end of the RTH it goes back to that position. So if the first line has not been successful, create another along this axis.

You couldn't be in sport mode otherwise it refuses the autonomous mission.

Ugh ... that is going to be hard to remember ... I think I might be able to extract that from the mission logs though let me go look. I know it has some early points which likely indicate the direction it was pointing.

If I am reading this right the yaw at takeoff was -121.6 ... which I think means the bearing was 238.4 ... and i it managed to clear all the trees on the way, and it traveled 90 seconds before finally "landing" there is a good chance the drone is on the edge of a pond in some wetlands >__<

Project Anavita Posted at 11-3 14:27
@Labroides

The short of it:

Sadly the FlightRecord log doesn't exist for this mission ... or any other ones I flew that day. In fact looking at the dates FlightRecord  hasn't recorded anything since 2020
That suggests that you weren't looking in the right place.
Were you using an Android or iOS device?

and the data you can export from the only .DAT file
The required file is a .txt file and it should be recorded for every flight.

The issue is where did it land.
I did follow your convoluted original explanation.
Your recorded flight data will show exactly where the drone came down and a lot more information about the flight.
Look again and if you can find that .txt file, I'll get the info you want from it.

I am trying to find where it might have "landed" by calculating the time it took to land, the likely speed of the drone, the topography and environment it was crossing over, and if the logs would indicate "landed" if it "crashed" instead.
It will be a lot simpler and more precise if you can find the flight data.

Labroides Posted at 11-3 17:23
Sadly the FlightRecord log doesn't exist for this mission ... or any other ones I flew that day. In fact looking at the dates FlightRecord  hasn't recorded anything since 2020
That suggests that you weren't looking in the right place.
Were you using an Android or iOS device?

I am using android yes.

I searched for all files modified in the last 7 days to find what was updated when I was looking for files that might be of use to me.

I then went to DJI folder and browsed through there when I found out about the FlightRecord before posting here might be a way to track down the drone.

My DJI > dji.go.v4 > FlightRecord has an item for my flight on Oct 26, the sub folder MCDatFlightRecords has Dat files for my Oct 31's missions but only seems to have recorded to beginning of those flights.

My DJI > com.pix4d.plugindji > FlightRecord folder only has updates up to Jun 2020 as previously stated.

If i search my whole device for DJIFlightRecord_2022 I only see three files. Two from July and one from Oct 26.

Project Anavita Posted at 11-3 17:41
I am using android yes.

I searched for all files modified in the last 7 days to find what was updated when I was looking for files that might be of use to me.

I've analysed hundreds of flight records, but I haven't used Pix4D so I've never had to find their files.
But following Airdata's instructions it sounds like they should be in there somewhere.
It would be very odd for them to stop recording flight data.

Project Anavita Posted at 11-3 17:41
I am using android yes.

I searched for all files modified in the last 7 days to find what was updated when I was looking for files that might be of use to me.

Is this any use to find the files?

https://community.pix4d.com/t/pi ... dji-go-4-logs/16667

I suspect there's a typo in the last line .. pilot instead of pliot

Labroides that was the key my friend.

Android > data > com.pix4d.plugindji > FlightRecord has the actual full record.

Including where it "landed"

Thanks man. No clue why my "search" feature on the android didn't bother to search the data folder >__<

And LV_Forestry you are correct ... it was pointing in the orientation it took off in. Sent the owner of the property a message of where to find the drone.

Project Anavita Posted at 11-3 18:25
Labroides that was the key my friend.

Android > data > com.pix4d.plugindji > FlightRecord has the actual full record.

Great result.
How far away from the launch point did it end up?

Likely hit its first tree at 504m, then another at 522m, and "landed" at 524m ... which I am "OK" with as 56m from that location on that trajectory is a body of water.

Drone recovered yesterday. Was 10ft in a tree in a landing position. A little dirty but still works

Project Anavita Posted at 11-5 04:00
Drone recovered yesterday. Was 10ft in a tree in a landing position. A little dirty but still works

Not even a damaged propeller?

LV_Forestry Posted at 11-5 23:39
Not even a damaged propeller?

Edges are a little dirty and likely have minor scratches which would increase drag but not chips or breaks.