Welcome to PhantomPilots.com

Sign up for a weekly email of the latest drone news & information

Aircraft lost control in follow me mode

Discussion in 'Pro/Adv Discussion' started by xbiron, Dec 9, 2015.

  1. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Hello guys,

    I've started a support ticket with DJI for this issue. But I was wondering if you guys know what happened ?

    Yesterday, I was trying the follow me mode on my P3P.
    I've started my flight in P-mode, then went to F-mode and waypoints (by mistake), i've canceled it and started the follow me mode.
    Everything worked great and the aircraft was following me. Then, at 10m03s of my flight, I've passed in P-Mode to change the position and altitude of the aircraft.
    Once he reached the wanted position, I've returned in F-mode and activated back the follow me mode at 11m01s.
    When I've clicked apply. The aircraft went "crazy" and went right full throttle, and finished his course in the trees. I tried to get back in P-mode, to take back control of the aircraft but it was too late. Fortunately, it just hit some branches, and was at low altitude (about 3m above the ground). So nothing got broken, and it works fine now. But I totally don't understand what happened.
    I wasn't touching the right stick for sure. He just decided to go towards the trees while I was standing in the street in front of the aircraft. Now i'm very scared to use this mode again.
    Here are the logs I got
    from the aircraft :https://dl.dropboxusercontent.com/u/18568770/P3P logs/FLY021.DAT
    and from my device :https://dl.dropboxusercontent.com/u/18568770/P3P logs/DJIFlightRecord_2015-12-05_[10-42-03].txt

    Thanks for your support.
    Xavier
     
    BudWalker and AgentOrange2015 like this.
  2. Oso

    Oso

    Joined:
    May 19, 2015
    Messages:
    2,539
    Likes Received:
    1,486
    Location:
    Sacramento CA
    Just guessing, but perhaps when you switched from P mode back to F mode, the P3 flew off towards the last set home point from when you were in F mode the first time.
     
  3. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    No it wasn't that. The home point was at the opposite way of where it goes.
    See the flight with some of my notes:
    Capture d’écran 2015-12-09 à 11.46.29.png
     
  4. Oso

    Oso

    Joined:
    May 19, 2015
    Messages:
    2,539
    Likes Received:
    1,486
    Location:
    Sacramento CA
    I see what you mean. It does seem to go in the opposite direction.

    Maybe one of the resident experts will take a look at the files you uploaded.
     
  5. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Ya I hope so… Thx man

    I've read some interference problem causing "flyaway" issues. Is it possible that my phone (pluged to the remote) did something like that ? or a strong wifi signal from a house near the street ?

    It's now 10 days I've sent my issue to Dji support. And I still have no answers. Is it normal?
     
    #5 xbiron, Dec 10, 2015
    Last edited by a moderator: Dec 18, 2015
  6. BudWalker

    Joined:
    Jun 24, 2015
    Messages:
    782
    Likes Received:
    230
    Location:
    Santa Margarita, Ca
    I didn't see your original post until just now. I'm looking at your .DAT and hope to have something informative to say in the next day or two.
     
    sbarton likes this.
  7. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
  8. BudWalker

    Joined:
    Jun 24, 2015
    Messages:
    782
    Likes Received:
    230
    Location:
    Santa Margarita, Ca
    I took a look at the .DAT. The following plots are from the point where Follow Me mode was engaged until the crash and fall to the ground

    FLY021.jpg
    The top plot is the velocity in the North direction. Since the P3 was moving south the velocity is negative. The third plot is accelerometer, the first spike is when the P3 hit the tree, the second spike is when it hit the ground. The second plot shows the control stick movements. It looks like you may have attempted to do a CSC, but it didn't have any effect. It is not evident from these plots, or anything else that I looked at, that the control stick inputs had any effect. The fourth plot shows the RC mode switch. It looks like you switched from F to A and then from A to P between the impact with the tree and the impact with the ground.

    Of particular interest was the information found in the text messages part of the log. Part of that is shown below

    Code:
    TickNo 452839:  37544 [Ctrl<4>] REQ_NAVI_MISSION NAVI_FOLLOW_MISSION ctrl_navi_follow_me_mission
    TickNo 453104:  37566 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453176:  37572 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453235:  37577 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453284:  37581 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453344:  37586 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453404:  37591 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453464:  37596 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453524:  37601 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453596:  37607 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453644:  37611 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453704:  37616 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453764:  37621 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453824:  37626 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 453883:  37631 [FOLLOW_ME]mission ERROR: target pos jump!
    TickNo 454303:  37666 [FOLLOW_ME]init gpitch -11.000,gdist offset -11.951,d2ph 83.469,d2prt 71.518
    TickNo 454303:  37666 [FOLLOW_ME] change to FOLLOW_MOVING
    TickNo 456355:  37837 [Ctrl<1>] REQ_RC_NORMAL ATTI ctrl_atti
    TickNo 456595:  37857 [FOLLOW_ME] change to FOLLOW_WAIT (lost 8,gps 3)
    TickNo 456595:  37857 [Ctrl<4>] REQ_NAVI_MISSION NAVI_FOLLOW_MISSION ctrl_navi_follow_me_mission
    TickNo 456715:  37867 [Ctrl<1>] REQ_RC_NORMAL ATTI ctrl_atti
    TickNo 456727:  37868 CTRL reset all by rc mode switch
    At TickNo 452839 the P3 had accepted the Follow Me mode command. TickNos 453104 through 453883 seem to indicate some kind of problem with this activation of the Follow Me mode. The first activation isn't shown here but is shown in the file FLY021.0X80.txt file attached.

    Mapping the plots above onto the message text is a little tricky, we're working on making that easier. The movement in the first plot started at about TickNo 454408, the control stick movements started at about TickNo 455125, the impact with the tree at TickNo 456130. The RC switch to A at 456727.

    My personal opinion is that there is a bug in the Follow Me mode that you managed to hit. The target pos jump messages hint at that. The software seems smart enough to know there is a problem and should be at least able to halt and not do anything. I couldn't find any indication of other problems

    I've attached some files that are the source of what's presented here. The FLY021.xls.txt is a Excel spreadsheet, the .txt was added so that it could be uploaded. You'll have to rename it by removing the .txt.

    You might also like this video that @Luap did with dashware.
     

    Attached Files:

    nparker13 likes this.
  9. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Wow thx man. So much data... Maybe I've tried to do something with the sticks later. I was panicking. But it's a while after the drone was going towards the trees that's sure. Do you think it's my phone s fault? Is it possible it didn't send the good gps informations? Or maybe the Dji go app was lagging on the phone?
     
  10. BudWalker

    Joined:
    Jun 24, 2015
    Messages:
    782
    Likes Received:
    230
    Location:
    Santa Margarita, Ca
    It's possible. I would assume that the DJI Go App had access to the number of satellites that the phone sees and, thus, some measure as to the accuracy of the location being sent by the phone. But still, that's a good question and I don't know the answer.

    If you want to look into it notice in the text file that I gave you the lines

    TickNo 204811: 16875 [FOLLOW_ME]init gpitch -35.000,gdist offset 27.480,d2ph 13.947,d2prt 41.426
    just as the first follow me starts and

    TickNo 454303: 37666 [FOLLOW_ME]init gpitch -11.000,gdist offset -11.951,d2ph 83.469,d2prt 71.518
    just as the second follow me starts

    I suspect that these lines somehow represent the location of the phone relative to the P3. If that's true and we could decode them then we'd know if the second is in error and causing the P3 to go the wrong way. I'd be glad to try and determine if these lines can tell us that if you could run some experiments. Basically, you'd need to have the phone and P3 in known locations and then initiate Follow Me. With a couple of these experiments we should be able to determine how to interpret those lines. Lemme know if you're interested; I'd want to make sure we're on the same page before you spend the effort to do the experiments.
     
  11. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Yes i will be interested. But how to do these experiments safely ? at low altitude with someone else holding the P3 legs ? Isn't it dangerous ?
    I'm not sur how you want me to determine the P3 and phone position.
    I understand that the 2 lines you quoted me are the points where follow me mode has been engaged but do you want me to understand something else with it? Because I have no idea what else they mean.
    And finally, one last thing i've forgotten to mention. When it happened, I was using my iPhone 5, because my iPad mini 4 is not GPS equipped. iPhone 5 is not in the recommended devices list of DJI. Do you think it could be the source of the issue ?

    Thx for your time

    PS: I still have no answers from DJI support. Do you think they will answer me one day?
     
  12. BudWalker

    Joined:
    Jun 24, 2015
    Messages:
    782
    Likes Received:
    230
    Location:
    Santa Margarita, Ca
    This isn't really going to do much for us. Even if that line shows the wrong info we don't know if it's because the phone sent the wrong location, or the app computed the numbers wrong. Let me think about this some. The location from the phone is probably somewhere and that's the info we want,
     
  13. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Ok sure. Thx man
     
  14. gegetrane

    Joined:
    Dec 12, 2015
    Messages:
    51
    Likes Received:
    9
    A similar thing happened to me the first time I tried Follow Me mode, and I'm quite sure it's my Nexus 7 losing its GPS signal that caused the bird to go wild. I now have to go walk with the Nexus while watching how it behaves on my GPS status app. Then only I'll try Follow Me mode again.
     
  15. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    I got an idea. In the DJI go app, it keeps a map with the route of the flight, including the airplane and the remote position. Here's the video of the route : https://dl.dropboxusercontent.com/u/18568770/P3P logs/route.mov

    Go to 2min18s, when I exited the first time of the follow me mode to P-mode, took back control of the airplane, placing it int he street at low altitude, then applied again follow me mode. Look at the remote position. It's still in the street, while the airplane went in the woods. But notice the airplace was rotating, still facing the remote like in a follow me mode correctly executed. So he knows the remote position was in the street. So weird.

    What do you think ? An iPhone 5 because not an A7 processor ?
    Thanks
     
  16. BudWalker

    Joined:
    Jun 24, 2015
    Messages:
    782
    Likes Received:
    230
    Location:
    Santa Margarita, Ca
    When I click that link I get an Error(404) not found coming from DropBox
     
  17. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Sorry, the video was uploading, now it's done, you can click again
     
  18. BudWalker

    Joined:
    Jun 24, 2015
    Messages:
    782
    Likes Received:
    230
    Location:
    Santa Margarita, Ca
    Can't really see what you're talking about at 2m18s. But, the behavior was clear at 11m2s. It could be that the phone new where it was and that's what you're seeing here. But, it's also possible that wasn't transmitted to the P3 correctly. I had located some other lat-lon in your.DAT that could be the P3's idea of where the RC is. I was going to look into that today. It would seem that DJI, Litchi, etc. all do the Follow Me differently. In the case of my Autopilot app that I used the RC's lat-lon info is done differently.
     
  19. xbiron

    Joined:
    Nov 25, 2015
    Messages:
    59
    Likes Received:
    4
    Sorry, I was speaking of 2m18s of the video time, not the flight time.
     
  20. MoOriginal

    Joined:
    Nov 24, 2015
    Messages:
    128
    Likes Received:
    23
    I had something similar with follow me mode - luckily I was high enough to catch it.
    Switched modes to adjust and take some pictures - sent it to follow me again it seemed to turn towards me and then shot off towards opposite direction...
    I was on a ipad air2 with good open signal and v1.5 firmware....