Mavic Mini Flyaway assistance wish

[KristofferR]

Hello!

I’m helping a friend by trying to recover his lost Mavic Mini.

In short, he appears to have done the usual mistake of flying in wind and panicking instead of just landing ASAP, eventually losing contact and then losing the drone.

In his defense, his house is behind a plateau. I live nearby and fly fixed wing mostly and during NW winds I avoid flying due to odd rotors and turbulence from that direction.

But still, I guess this will be a recurring theme with the Minis and inexperienced pilots flying downwind.

Wind acc to the nearest lighthouse station was 9 m/s at 323 (NW) so I did some hand-calculation attempts at estimating the possible landing zone, assuming the Mini is trying to RTH after 11 sec and 51% battery from his screendump and assuming drift speed 1,5 m/s, but the area is simply too big and my calculations too inexact.

Now I got the logs synched to
Airdata. They state a slightly different 293 deg wind. I hope someone here could assist me to improve the landing precision. There are several big fields in the area as well as dense forest, with some luck it could be recovered.

Airdata log:

 

[Lurch003]

Another "helping a friend" ?!!

I hope you find your drone

 

[KristofferR]

It is a friend! I have flown fixed since the 80's with some forays in multirotor and fpv.

If I messed up I would own up to it.

The day he lost his drone I was happily slope-soaring the NW wind....

KristofferR's blog - RC Groups

RC Groups - the most active Radio Control model community: electric and fuel rc airplanes,rc helis,rc boats and rc cars. Features discussion forums, blogs, videos and classifieds.

www.rcgroups.com

 

[Meta4]

In short, he appears to have done the usual mistake of flying in wind and panicking instead of just landing ASAP, eventually losing contact and then losing the drone.

Wind acc to the nearest lighthouse station was 9 m/s at 323 (NW) so I did some hand-calculation attempts at estimating the possible landing zone, assuming the Mini is trying to RTH after 11 sec and 51% battery from his screendump and assuming drift speed 1,5 m/s, but the area is simply too big and my calculations too inexact.

Your friend put the Mini in Tripod mode 7 seconds after launching and left it in Tripod Mode until 5 seconds before signal was lost.
He then took the drone up to 80 metres where the wind was stronger than what he was experiencing on the ground.
Tripod mode has a top speed of 4 metres/sec.
In Tripod Mode with the right stick pushed full forward and the Mini pointing towards home, the Mini was being blown further away at 4 metres/sec.

Had he taken the drone out of Tripod Mode, and descended out of the strong upper level wind .. he could have easily brought it home.

On losing signal, the Mini would have attempted to come home (still 80 metres up) against the wind, but being 570 metres away when it started fighting the wind, it was too late.

 

[KristofferR]

Thanks Meta4.

My best estimate from what I read here is 1% battery per 10 seconds.
Wind speed of 9m/s

RTH activation after 11 s means 11*9 = 100 m "free drift"

50% battery and assuming landing at 10% battery = 40% * 10 seconds = 400 seconds before land.

Then would it drift at 9 - 7 = 2 m/s with a slant towards the RTH point. 800m away.

Total 900m SE

I don't know the descent speed and time

I have seen some better numbers and graphs presented but don't know the data.

 

[slup]

In this flight it was several things occurring that makes it really hard to predict a possible touch down ... as seen in the log your friend first of all apparently didn't check the map where the HP had been recorded, it wasn't from where he took off. Then he strangely didn't yaw anything, didn't try to point the AC back to him at all ... instead he probably confused used a lot of un-logical negative elevator (backing the AC) & positive aileron inputs instead. Then as said earlier he stayed in Tripod mode nearly the whole flight & ascended instead of descended. Further more the payload mode was activated which indicates flying with the prop guards ... that also makes the Mini's possibility to fight wind very minimal.

The fact that he not once applied inputs towards either the recorded HP or from where he took off ... instead in other directions makes it impossible to interpolate the following drift speed when the connections was lost & the failsafe RTH kicked in. Below the telemetries in the end of the flight ... interpolating the battery % gives further 365,3sec of flight until autolanding would kick in from an height of 80m ... the Mini have 1,8m/s in descend speed.



The probable "Blow away" path would not have been straight ... this due to the side wind & HP location, it looked most probably like the dotted yellow line below, where that bowed line ended or how long the Mini had battery to follow it is unknown.

 

[KristofferR]

Many thanks for those numbers! its a much better estimation than mine and also a fair bit closer.

You are correct about the prop guards. When he showed it, they were mounted. I did not understand the payload message but then it makes total sense. Also, I was curious about the weird home point in the log and what the tripod mode was for.

We will go to the farm and ask permission to search their lands. It's all pastures and
still short spring grass. The surrounding terrain is extremely rugged and inaccessible. There is recovery potential if it went down somewhere on the fields.

I also promise to give him a proper flight education, regardless of the outcome.

 

[slup]

Many thanks for those numbers! its a much better estimation than mine and also a fair bit closer.

You are correct about the prop guards. When he showed it, they were mounted. I did not understand the payload message but then it makes total sense. Also, I was curious about the weird home point in the log and what the tripod mode was for.

We will go to the farm and ask permission to search their lands. It's all pastures and
still short spring grass. The surrounding terrain is extremely rugged and inaccessible. There is recovery potential if it went down somewhere on the fields.

I also promise to give him a proper flight education, regardless of the outcome.

Just want to point out that the yellow dotted line isn't any thing calculated it's just a visual example of that the path most probably was curved not straight ... no need to search that one off.

Theoretically if a to high wind velocity is affecting the AC from the side while the AC is trying to make progress 90 degrees towards it, the path should be curved. If the battery lasts, this will continue until the AC have direct head wind whilst trying to make the way to HP ... but then if the wind is to strong the flight path will bow back & then blowing away in the wind direction (path kind of equal to an S-shape). Where the AC is located when the battery reaches auto land is impossible to predict due to that the path is unknown ... it's not as simple as calculating a head wind drift speed together with remaining battery & land height to come to a probable spot.

 

[sar104]

Just want to point out that the yellow dotted line isn't any thing calculated it's just a visual example of that the path most probably was curved not straight ... no need to search that one off.

Theoretically if a to high wind velocity is affecting the AC from the side while the AC is trying to make progress 90 degrees towards it, the path should be curved. If the battery lasts, this will continue until the AC have direct head wind whilst trying to make the way to HP ... but then if the wind is to strong the flight path will bow back & then blowing away in the wind direction (path kind of equal to an S-shape). Where the AC is located when the battery reaches auto land is impossible to predict due to that the path is unknown ... it's not as simple as calculating a head wind drift speed together with remaining battery & land height to come to a probable spot.

I think that you may have made an incorrect assumption here. In any of the P modes the primary FC goal is positional hold, followed by progress in a given direction. It will use tilt to hold position and then use any additional available tilt to make progress. When flying towards a target it dynamically uses roll to correct for sideways drift. However, if the wind speed exceeds the maximum airspeed - i.e. full tilt doesn't achieve position hold - then there is no further tilt available.

As a result we don't see curved tracks in blowaway events - the aircraft always drifts approximately downwind even if that is not directly away from the homepoint. Applying the usual calculations to this event gives 355 seconds to autoland start. The computed wind speed was 13.5 m/s and the Mini can only manage 8 m/s, so it should have drifted at around 5.5 m/s. That leaves 36 seconds to land based on the slightly higher ground level at the distant location:



On the map that puts the estimated landing point at the yellow pin, which is around 2 km away. I added 5° directional error bounds and 5% distance uncertainty to get the green box.

 

[slup]

... In any of the P modes the primary FC goal is positional hold, followed by progress in a given direction...

That's good learning, thank's @sar104 ... & accordingly, I stand corrected

 

[KristofferR]

Thanks sar104! I thought long and hard about the S-curve and decided that it does not seem plausible, and assumed the drone would head towards the home site as best as it could since it is a free-floating object in the wind - it will just point towards home direction all the time at the rated speed and eventually it should line up straight downwind of the Home point and keep on in a straight direction. It should be possible to calculate the landing point with good precision.

With that assumption, I created a simple Python program that updates position and angle every second, added a little offset for the 11 seconds of response to lost signal and ran it for the 356 seconds with the indata of wind angle, wind speed and start position relative to home.

The result was 1939 m east and 808 m south which viewed on the map is pretty much exactly what sar104 estimated.

The hardest variables are wind direction and wind strength. Since the resulting speed vector in a headwind is the (wind speed - drone speed) even a small difference in the estimated wind speed will make a big difference to the estimate.

 

[slup]

Thanks sar104! I thought long and hard about the S-curve and decided that it does not seem plausible, and assumed the drone would head towards the home site as best as it could since it is a free-floating object in the wind - it will just point towards home direction all the time at the rated speed and eventually it should line up straight downwind of the Home point and keep on in a straight direction. It should be possible to calculate the landing point with good precision.

With that assumption, I created a simple Python program that updates position and angle every second, added a little offset for the 11 seconds of response to lost signal and ran it for the 356 seconds with the indata of wind angle, wind speed and start position relative to home.

The result was 1939 m east and 808 m south which viewed on the map is pretty much exactly what sar104 estimated.

The hardest variables are wind direction and wind strength. Since the resulting speed vector in a headwind is the (wind speed - drone speed) even a small difference in the estimated wind speed will make a big difference to the estimate.

All comes down to which strategy DJI have chosen ... think that if position hold haven't had absolute priority, instead used all available tilt (combination of pitch & roll) equal parts to the wind direction & then also into the HP direction the path had been curved ...

But as pointed out that's not the strategy they chose, the tilt direction here was full into the wind for position hold leading to drift direction equal wind direction.

With that settled, one parameter is fixed ... then the drift speed, the only thing that's available is knowledge about that the Mini can have a max speed of 8m/s in P-mode ... & the estimated wind speeds from for instance Airdata.com here below ... Good luck!

*EDIT*
A bit regarding those 11sec until failsafe RTH initiates ... no need to treat that in a special way, as position hold is in priority the Mini will drift during that period as much as after as the tilt angle already is maxed out, it will only yaw towards HP nothing more.

 

[droneloverair]

When in tripod mode, would enabling sports mode switch disables immediatly the tripod mode?

 

[KristofferR]

I decided to go for the estimated speed - the weather station at sea says 9 m/s but the same wind was yesterday and the speed is definitely higher over the hills 2km inland where this happened. I spent some time yesterday evening checking out the area with no luck.

Lots of tall grass, reeds, sheep and horses plus lots of machinery and farm-related stuff provide lots of hiding places for a tiny drone. Still, also means there is a decent chance of discovery by the locals. Name, phone number and a finders fee were written on the drone. It's a nice place and several trails around so good for a few more walks.

 

[sar104]

I decided to go for the estimated speed - the weather station at sea says 9 m/s but the same wind was yesterday and the speed is definitely higher over the hills 2km inland where this happened. I spent some time yesterday evening checking out the area with no luck.

Lots of tall grass, reeds, sheep and horses plus lots of machinery and farm-related stuff provide lots of hiding places for a tiny drone. Still, also means there is a decent chance of discovery by the locals. Name, phone number and a finders fee were written on the drone. It's a nice place and several trails around so good for a few more walks.

What do you mean by the "estimated speed"?

 

[KristofferR]

Estimated speed - I should have written “the wind speed as estimated by the drone/Airdata”.