Sean-bumble-bee Posted at 4-12 11:25
Actually I think in this post you are possibly in error. I would be surprised if much bounce was due to elastic rebound (think bouncing ball) or residual speed/energy left in the props/motor combination. I wouldn't mind a tiny bet that the drone's control system belatedly had some sort of reaction and revved the drone's motors.
It was/is one reason I was hoping the DAT would be available.

Do you think that it reacted to the wall? I have only heard of one person (the thread i linked to earlier in the thread) that mention that it moves away when OA detects during landing, its not happened to me yet and i hand catch all the time when not at home, perhaps its because I only use brake on detection?

Another thing is that I have never seen my Mini 3 Pro land that fast before which is why i said what i did.

A harder landing than it normal would inevitably cause some bounce and as its come down too hard that means that the props are not as slow as they should be hence the move forward. I am not saying that its considerably faster than it should be, its just faster, this is why someone near the beginning thought the video was sped up.

The above can be tested but its risky as it would mean having to recreate the OPs landing, i might be able to try this myself as i have a low pallet fence but it would need to be a calm day.

Another wild thought was, how wide is the downward sensor spread?
During the landing, was the higher ground detected, then at the last second it picked up that the ground is actually lower than it initially computed for?  Would there be time to add this to the log?

Then there is the auto cam pitch to 0º in my opinion, that was far too late but to be fair, i do not use RTH landing with the Mini's so i am basing that thought on my P4P when i have auto landed with that, so there's another thing for me to test but it might not be until Saturday but more likely Sunday unless anyone else has lower winds than before hand, I would like to see videos too as i will record mine.

"Reacted to the wall", in what sense?
I am not in a position to comment on precisely how the mini 3 reacts to OA input during descents and perhaps the 'bounce's'  'climb' other than to relate it to my experience with the M2P and that was a nightmare when low as it got 'stuck'.  I do not use the M2P's OA.

Your point about the motor speeds during a descent raises an interesting question, what are they?
I can only think that strictly speaking they must be very near to hover speeds otherwise the drone would accelerate during its descent. ( It's similar to the "what do you weigh in a lift" thing, your weight (as felt by your feet) changes from normal only whilst the lift is acceleration, when the lift is moving at a steady speed your weight is 'normal'. )
If you get a windless day it might be a worthwhile experiment to send the drone up high and then do an automated descent to get the various descent speeds that JJB mentioned then download the DAT from the phone and process it with CsvView to see what the motor speeds were during the various descent speed phases.

I would not be inclined to try to recreate the OP's descent, I think the only way you could do that is to mask off the downward looking sensors and might leave your drone descending at speed very close to the ground. JJB mentioned that in test he made the descent speed started to decrease at around 50m which I am pretty sure is above VPS range and that would imply that the barometer reading governs that threshold. With my none mini 3 drones I get the feeling that VPS governs the descent speed but kick in at lower heights.

With regards to bounce, I am just left witha) the feeling that there isn't the elasticity in a drone's shell nor the descent speed to produce bouncing ball 'bounce' that we saw and
b) the suspicion drone's control system is responsible for the bounce.

With regards to the camera tilt, initially I too thought the camera's levelling was "late" but thinking about it, at what height does a mini 3 normally 'level' the camera and how quickly does it do it? I don't know.
The question then becomes, in this flight, at what height did it start to do so, I will perhaps have a look later today. Given the descent speed it may be that it stared to level the camera at the normal height but was just 'slow' in relation to that descent speed.

Sean-bumble-bee Posted at 4-12 21:00
"Reacted to the wall", in what sense?
I am not in a position to comment on precisely how the mini 3 reacts to OA input during descents and perhaps the 'bounce's'  'climb' other than to relate it to my experience with the M2P and that was a nightmare when low as it got 'stuck'.  I do not use the M2P's OA.

I was wondering if you thought that it reacted via the OA.

I initially thought about the taping up of the bottom sensors and was about to type it but then I also thought that that wouldnt be any good as it wouldnt be recreating the decent to see if the close quarters to the wall was the cause.

Going by the log, the auto cam pitch was at 0m at 6:54:0
I think that the cam returns to 0º right before its final slowdown to actually land.
I have watched this video so many times and i do not see any evidence of the final slow down prior to actually landing, it should have been before the cam pitch, i am sure of it although i cannot say definitively as its been a too many months since i have flown the P4P and use RTH through to auto-land.

Sean-bumble-bee Posted at 4-12 21:00
"Reacted to the wall", in what sense?
I am not in a position to comment on precisely how the mini 3 reacts to OA input during descents and perhaps the 'bounce's'  'climb' other than to relate it to my experience with the M2P and that was a nightmare when low as it got 'stuck'.  I do not use the M2P's OA.

Hi Sean,

See my excel file i posted previous with a link to it for a normal descent of a MINI3PRO

Yes, baro height regulates downspeed below a threshold height/Normal from high heights rate of decsent is 3 m/s, reduced to lower value from approx 20 meter, within IR sensor range more reduced to 0.1 m/s in a kind of linear reducing.
So first baro height info is used, within IR range this sensor is used to control down speed *
* ofcourse also IR sensor to do this, otherwise all landings where baro height shows a bog plus or big minus height at landon would not go a normal.

Checked few MINI3PRO flight ; the autofunction to level camera during landing starts between 2.0 and 1.8 meter IR height and takes from 90 degrees down to zero less than 1 second. So always level < 1 to 1.5 meter IR height.
cheers
JJB

FYI, I had the same issue but was able to stop auto landing just in time.

TrippleDelta Posted at 4-13 00:20
FYI, I had the same issue but was able to stop auto landing just in time.

can you post the logs? The more logs of similar issues the better.

Bashy Posted at 4-12 23:08
I was wondering if you thought that it reacted via the OA.

I initially thought about the taping up of the bottom sensors and was about to type it but then I also thought that that wouldnt be any good as it wouldnt be recreating the decent to see if the close quarters to the wall was the cause.

No, a pure guess, though it seems to be a bit logical to me, is that the initial vertical portion of the bounce/s might be a belated Landing Protection reaction, same as you get if you put your hand under the sensors, coupled with continued downstick by the OP, I'll need to check that later, on the wrong device at the moment.

With regards to OA I have noticed that the manual details how OA will work during RTH climb to height (brake and move horizontally away from the object) and during the horizontal flight home ( brake and climb). It also makes mention of the drone climbing during the horizontal flight home if it senses something underneath the drone. BUT, that I can see, it makes no mention of what happens if the drone senses an object in front of or behind the drone during the descent, leaving me puzzled as to whether or not it reacts to such an object. I am also curious as to whether or not the response to something close to the bottom of the drone during the horizontal flight home is anything more than a "hand under the drone" response or a just 'marketing ploy' to "talk up" the "hand under the drone thing".
Then there is the section on APAS which baffles me lol and makes me wonder if it is duplication of stuff already described elsewhere in the manual. From memory I had the same thoughts about the APAS section in the Mavic 2 manual.


I think if I owned a Mini 3 Pro there would be a lot of experimental flights, to clarify my understanding of the manual, before I sent it out on a proper flight.
Given the manual's reference to the drone moving to avoid an obstacle during the climb to RTH height, I am left wondering if it does the same thing during ANY climb and if so could the forward movement during the bouncing be a response, during each upward portion of the 'bounces' to the low wall behind the drone.

I think it would be a really good idea, but unlikely to happen, if one of the DJI engineers weighed into this thread and explained what we are seeing, this thread seems to have raised an awful lot of questions.

JJB* Posted at 4-12 23:16
Hi Sean,

See my excel file i posted previous with a link to it for a normal descent of a MINI3PRO

I will look later.

Sean-bumble-bee Posted at 4-13 01:45
No, a pure guess, though it seems to be a bit logical to me, is that the initial vertical portion of the bounce/s might be a belated Landing Protection reaction, same as you get if you put your hand under the sensors, coupled with continued downstick by the OP, I'll need to check that later, on the wrong device at the moment.

With regards to OA I have noticed that the manual details how OA will work during RTH climb to height (brake and move horizontally away from the object) and during the horizontal flight home ( brake and climb). It also makes mention of the drone climbing during the horizontal flight home if it senses something underneath the drone. BUT, that I can see, it makes no mention of what happens if the drone senses an object in front of or behind the drone during the descent, leaving me puzzled as to whether or not it reacts to such an object. I am also curious as to whether or not the response to something close to the bottom of the drone during the horizontal flight home is anything more than a "hand under the drone" response or a just 'marketing ploy' to "talk up" the "hand under the drone thing".

Next week lots of sun and no wind in my county

Time to do some test flights and video them. Will autoland my drones just behind and very close to a  brick wall....  ;-)

In an autolanding the OA sensors does see object close, but drone will not move away from an object.

Hand under a flying drone will raise the drone.
This behaviour is not active during during last part of landing, obvious  otherwise drone could never land.

cheers
JJB

JJB* Posted at 4-13 02:57
Next week lots of sun and no wind in my county

Time to do some test flights and video them. Will autoland my drones just behind and very close to a  brick wall....  ;-)

Cheers.
I was not questioning whether to not the "hand under the drone" thing is active once the drone has started the 'below 50cm' final descent. As you say it couldn't be otherwise the drone wouldn't land, in fact I think I have put my hand under a mini that was in that phase and hand caught it.

Sean-bumble-bee Posted at 4-13 03:13
Cheers.
I was not questioning whether to not the "hand under the drone" thing is active once the drone has started the 'below 50cm' final descent. As you say it couldn't be otherwise the drone wouldn't land, in fact I think I have put my hand under a mini that was in that phase and hand caught it.

Theres 2 ways that i catch it...

1) hold out my hand and bring it down, it senses hand and it slows like it should for a gentle catch
2) bring it down a bit higher than head, put hand under it to force it to go up whilst holding down the stick to catch it.

Both times i have it in sports mode to shut off the OA, after matey had that issue i didn't want to risk it.

Looks like we will all be doing some testing in the near future, by rights we should all see the same results.

1 question, if i send it straight up, then hit RTH am i right in thinking that it will auto-land in place?
I want to mimic the OP's as best i can so RTH will be apart of it...

Bashy Posted at 4-13 03:27
Theres 2 ways that i catch it...

1) hold out my hand and bring it down, it senses hand and it slows like it should for a gentle catch

If it is within 5 horizontal metres of the home point an RTH should make it descend where it is.
" The aircraft will land immediately if it is less than 5 m from the Home Point when RTH begins."

Sean-bumble-bee Posted at 4-13 03:38
If it is within 5 horizontal metres of the home point an RTH should make it descend where it is.
" The aircraft will land immediately if it is less than 5 m from the Home Point when RTH begins."

brill, thanks, makes things a little safer as long as it comes down on the spot

Hi,

Add some code to FRAP, now cam pitch data during landing is visible.
Nice to compare logs for this auto-cam-to-zero moment+duration.

cheers
JJB

JJB* Posted at 4-13 05:09
Hi,

Add some code to FRAP, now cam pitch data during landing is visible.

Thanks for that, good to know, so it is quite low then, last second so to speak, was i as low as in the OP's video, perhaps because it came in hot made it look later than it should have been?

Bashy Posted at 4-13 06:36
Thanks for that, good to know, so it is quite low then, last second so to speak, was i as low as in the OP's video, perhaps because it came in hot made it look later than it should have been?

OH, my bad, thats IS the OP's log still, lol i thought you'd done a test...

excel file in post 69   shows normal landing for mni3pro, see also the difference in rate of descent  :-)

next week fly drones test week...  ;-)

Bashy Posted at 4-13 03:27
Theres 2 ways that i catch it...

1) hold out my hand and bring it down, it senses hand and it slows like it should for a gentle catch

Bashy, I would turn OA off in settings. Too easy for something to go wrong in Sport. But, that's just me.

JJB* Posted at 4-13 06:53
excel file in post 69   shows normal landing for mni3pro, see also the difference in rate of descent  :-)

next week fly drones test week...  ;-)

Sorry, i should have stuck with my initial thoughts, changed it back now lol

Mobilehomer Posted at 4-13 08:08
Bashy, I would turn OA off in settings. Too easy for something to go wrong in Sport. But, that's just me.

For hand catching it doesn't make much difference, not for me anyway as I never change the drone's position once I start the descent, there is no need to change the drone's position to be fair as I myself move if needed and the downward sensors still work as they should.

There will be times when others will need to change the drone's position such as on a boat for example so for them N mode with OA off would be preferable.

Hi

If you like , check this

https://forum.dji.com/forum.php? ... p;extra=#pid3024255

cheers
JJB

JJB* Posted at 4-14 11:32
Hi

If you like , check this

Seen that, thanks and yep, it confirms that the OP's landing was too fast.

We're still left with, why was the OP's landing too fast? we will probably never get that answer unless they send it in for analysis.

JJB* Posted at 4-10 06:34
Mode changes are not ontime after starting to fly.

Start a fllight with bypass ON , see that after take-off this is not in the log directly, but after short moment change to the correct setting.

Hey JJB have you had a look at the battery temperatures??????????????????????????? Are they normal?

Sean-bumble-bee Posted at 4-19 10:05
Hey JJB have you had a look at the battery temperatures??????????????????????????? Are they normal?

Yes,  high at start, cooling down in flight  (dropped 10 degrees in flight...)

So if batt temperature had a negative effect on this flight than most likely at start of the flight.
Plus no battery warnings in the log.  

cheers
JJB

The same thing happened to me a few days ago. Latest firmware installed, flown in cold temperatures and RTH triggered. The drone came back and descended at 5 m/s. Before I could react, she crashed into a meadow. The battery was at 17 degrees Celsius and was still about half full. When the crash occurred, the log showed the message that the rotors were blocking. In addition, a height of 49 meters was issued. In fact, the drone had already reached the ground. I have no clue whats the cause of this. Never had any problems with RTH. Fortunately, the drone remained undamaged.

Paul Dash Posted at 12-4 11:11
The same thing happened to me a few days ago. Latest firmware installed, flown in cold temperatures and RTH triggered. The drone came back and descended at 5 m/s. Before I could react, she crashed into a meadow. The battery was at 17 degrees Celsius and was still about half full. When the crash occurred, the log showed the message that the rotors were blocking. In addition, a height of 49 meters was issued. In fact, the drone had already reached the ground. I have no clue whats the cause of this. never had any problems with RTH. Fortunately, the drone remained undamaged.

Perhaps you could post the flight log? I.e. upload the screen device .txt flight log to https://www.phantomhelp.com/logviewer/upload/
and post the resulting URL here.
17C is cold for a battery isn't it? I think I get warnings around or not far below that temperature with my drones (not Mini 3 ). Grass probably jammed a motor so that warning is not unexpected.

Paul Dash Posted at 12-4 11:11
The same thing happened to me a few days ago. Latest firmware installed, flown in cold temperatures and RTH triggered. The drone came back and descended at 5 m/s. Before I could react, she crashed into a meadow. The battery was at 17 degrees Celsius and was still about half full. When the crash occurred, the log showed the message that the rotors were blocking. In addition, a height of 49 meters was issued. In fact, the drone had already reached the ground. I have no clue whats the cause of this. never had any problems with RTH. Fortunately, the drone remained undamaged.

If you have not yet synced the flight data with DJI perhaps you could also upload the screen device .DAT flight log to a file hosting website, make the page public and post a link here.

https://www.dropbox.com/scl/fo/blpn19xil8vwjo9i8ymfv/h?rlkey=a6gptbrw5osufb4h1tb44rhbl&dl=0

Hope that it  works. Seems that the IMU- and VPS-Sensors had different data…
I had no battery-warning while flying. The jammed rotors occoured after the crash because of the grass.

Paul Dash Posted at 12-4 12:08
https://www.dropbox.com/scl/fo/blpn19xil8vwjo9i8ymfv/h?rlkey=a6gptbrw5osufb4h1tb44rhbl&dl=0

Hope that it  works. Seems that the IMU- and VPS-Sensors had different data…

You'd do better to share the URL of the phantomhelp page rather than screen captures of its scrolled summary.
If you don't want to make the page public PM the URL to those who you think can decypher it.

Any luck with the DAT ?

Since you seem to have landed near the take off point I assume the IMU height is flawed and that it should have been near zero.
What was the air temperature and where you in fog or rain etc.?

Yes, I also assume that it has to do with a faulty imu. But how can this happen? It was around 0 degrees Celsius, it was cloudy and therefore a bit dark. no rain, snow or fog.

Paul Dash Posted at 12-4 12:08
https://www.dropbox.com/scl/fo/blpn19xil8vwjo9i8ymfv/h?rlkey=a6gptbrw5osufb4h1tb44rhbl&dl=0

Hope that it  works. Seems that the IMU- and VPS-Sensors had different data…

Seems that the IMU- and VPS-Sensors had different data…
That's obvious and a bit interesting, but it really doesn't matter for landing anyway.
The drone doesn't have to land where it launched.
It can land somewhere that's a hundred feet higher or lower than where you launched without any problem.
And the VPS is obviously working to help slow the descent before launding anyway.