Well, after having my Mavic for a few days, and after a dozen successful flights, it decided to spaz out on me the other night. The end result was a pretty bad crash into a tree, and a substantially damaged drone. I actually feel fortunate that it was just a tree that it hit. The Mavic flew completely out of my field of view, and I was really worried about where the hell it would end up, or what/who it might crash into. I haven't submitted a report to DJI, yet, but here's my account of what happened:
I launched the drone from the balcony of my house. By now I had done this several times. I had become quite confident with my piloting skills, and launching from my balcony was not an issue. Immediately after takeoff, I gave it a good amount of thrust and forward stick, sending it up and away from my house. I didn't fly it out too far. I paused for a while, spun the drone around to face back at us, and then flew it in closer to the house.
When it was a reasonable distance from us, I wanted to pause and let it hover, and then adjust the angle, focus etc to do a bit of filming of all of us on the balcony. However, the drone started moving in a way that didn't seem to correspond with my control inputs. At first I thought it was just "lag" after some fairly heavy inputs. However, it soon moved very quickly, and flew over our heads and over the roof of my house, completely out of view. This all happened so quickly that I barely even remember what was happening. I know that I soon looked down at my screen to try to figure out what it was doing, and what I saw looked like it was spinning. (As it turns out, it wasn't spinning, but flying sideways very quickly).
Very soon after, the remote lost all connection, and I had no idea where the drone was. One of the last things I saw on my screen was "Motor obstructed," which gave me the clue that eventually helped me find and recover the drone. Based on the direction I saw it heading, I looked for likely targets, and there were a few trees that were likely candidates. Sure enough, I found the drone on the road below a very large tree, quite badly damaged. The battery had separated completely (explaining the complete signal loss) but was still in-tact and turned on, there's a bunch of wires hanging out of the camera/gimbal, three props were broken or nicked, and one of the rear legs is snapped.
I appear to have all the flight log data, which I guess I'll need to send to DJI for analysis. However, my own initial analysis seems to indicate a similar "sticky left stick" issue that I've seen reported by a few other users, most notably this incident: http://forum.dji.com/thread-67922-1-1.html
At first I thought perhaps the drone had entered Return to Home (RTH) mode (perhaps a problem with the app/wifi/controller), and then maybe due to a GPS signal issue, it may have flown to the wrong location. However, when I looked at the video of the flight, I can see that the drone was actually flying sideways up until it hit the tree. I know that the RTH function always points the drone in the direction that it needs to go, so this wouldn't make sense. Also, I never got any warning or anything on my controller or app about it entering RTH. The drone simply flew off on its own.
Now, in my case, I had customized my controller configuration so that my left stick controls left/right movement instead of the default of rotation. So, a left stick stuck to the left in my case would cause the drone to fly sideways, rather than spin or arc. When I reviewed the flight log in the Go app, I can see at some point there's some left stick movement that I'm quite sure I didn't make. At one point just before the crash, the left stick fluctuated up and down between 28, 50, and 100% leftward direction. I'm almost completley sure I didn't make those control inputs, because I know that the best thing to do when the drone is out of sight or out of control is to let go of the sticks and let it hover or RTH. As soon as I saw it shoot over our heads and out of sight, I let go of the sticks, hoping it would hover. After that, I honestly don't think I made many control inputs, as they would have been completely blind inputs, which wouldn't make any sense. Strangely, if I view the controller log and try to roughly sync it with the video recording, the controller log does not show any inputs at all while the drone is clearly flying sideways for about 10 seconds before hitting the tree. The controller log even shows horizontal speeds of up to 7.6 m/s with no controller inputs.
After the incident, I thought at first that I may have made some sort of piloting error. Like maybe not correctly understanding what would happen in the case of weak GPS signal, or something else. But after seeing a few other reports of the sticky left stick, I'm starting to think it wasn't my fault afterall.
Since I know that a few people are compiling information on these incidents, here's a few other pieces of information:
This was my 14th flight with the Mavic (according to the Go app), and I don't remember having a hint of a problem before this crash.
The Mavic, controller, etc were all fully updated.
My phone is a Galaxy S7, fully updated.
I had flown the drone in the same area (my garden, also launched from the balcony) just a few hours before, without any issues.
I had not done a compass calibration before this flight, but the drone seemed to be working fine at the beginning of the flight, properly responding to my inputs.
I've never done an IMU calibration, as I understand that should only be done when the app tells me to do it.
I guess my next step will be to open a ticket with DJI support, and then see what happens. I'll try to keep this thread updated with as much information as I can.
I don't like to sound alarmist, but this experience really has scared me a little bit. My incident could have been so much worse if it had crashed into a person. The fact that this happened out of the blue without any warning is quite worrying.
One thing I notice: The takeoff location that I can see in the RC log actually looks to be about 50m off from where the fly actually started.
UPDATE #2 - I've now put videos of the crash and the aftermath onto Youtube:
The crash:
I think it was around the 20s mark in the video where the drone started to act on its own.
In several Forums I saw discussions about possible influence of Magnets or Metals to the Remote as possible reason for the stick problem.
Did you have a watch, a ring, a metal-table or any other piece of metal or magnet at or near the Remote when the probem occured?
Can you upload your flight log here, you can upload to healthydrones.com, so people can see.
You can also upload your flight record first download to Dropbox and post link here,
These are helpful to other flyers here to get some understanding how these things happen..
Wow, what a complete disaster! One mavic issue after another and they just come on out of the blue. So tell me again why I am waiting online for one at thIs point?
Sorry for your crash, please contact DJI support to have your aircraft/RC sent in.
Also what is your email and I will look at your flight later today. I will also email you a file so you can download all RC data. And please download the DAT and VSS file of that flight and upload it to Dropbox and provide the link.
Complete disaster, there's thousands of Mavics out there and we have only a tiny handful of users who have experienced this and R&D is working on it.
The email is onlinesales@dji.com to cancel your order.
And I'm sure that given time, more and more of these issues will continue to pop up to the ones that are out there until they get a handle on what's causing it , until then, how is it safe to fly? And to cancel my order, I send an email, wait until they can get to me, whenever that might be, then wait another 4 to 5 days to get my refund?
Nees Posted at 2016-11-6 07:20
You say: I've never done an IMU calibration, as I understand that should only be done when the app t ...
You do NOT Need IMU calibration. Only if Apps is asking. It as been discussed so many time before. Maybe it's why so many people having troubles. doing things you shouldn't do in first place.
Yep, I get the feeling allot of people are confusing the compass for the IMU and not realizing that these are two separate instruments.
So what its sounds like is that people are calibrating thier IMUs before each flight, out in the field on uneven terrain..
I've had 4 DJI quad rotors now and i don't think I've calibrated the IMUs on any of them, Ever. Compass only calibrated once out of the box and then if I've taken them on an overseas trip.
Seem like it might be a good idea to remove the IMU callibration option from the app..
Make people do it via the PC/Mac Assistant only when absolutely needed..?
R&D has completed testing and they iWatch with magnetic band has minimal effect on the RC and only when it is in contact with the case. If you are holding the RC normally it does not effect operation. If you having an issue with what you feel is a sticking stick then please report it to DJI.com/support. They can assist further. OP I will report your case but please provide your flight data logs to them for analysis.
Yep lot of things are confusing. Like don't touch your control stick after you intialize RTH. you may lose the craft. And so many are still doing it. Most of this staff are in the manual.
Maybe DJI should do a real DO and DON'T DO list.
Like a proper one by DJI engineers or people who really know how the mavic really work, not by people who believe "maybe it's working like this way".
By the way I am in Australia and we are still waiting...
Jeromeo Posted at 2016-11-6 13:55
Yep lot of things are confusing. Like don't touch your control stick after you intialize RTH. you m ...
agreed. Misinformation spreads like wildfire on this forum.
Waiting for mine right along with ya.
Heading over to Australia on the 21st for 3 weeks. knowing my luck though, the Mavic will show up the day after i leave.. Fingers crossed they ship it in the next 7 days
Hi.
Tech question for DJI. If the issue of Mavics moving on their own is due to a glitch in the joystick behaviour, how the logs will detect if this glitch is due to an HW defect or to an user action ?
Jeromeo Posted at 2016-11-6 08:55
Yep lot of things are confusing. Like don't touch your control stick after you intialize RTH. you m ...
No.
We dont need workarounds but a bugfixed Firmware.
(at keast a fix for that fatal one).
And all this mumbo-jumbo about Compass failure and wrong calibration:
There have been much simpler and less sensor equipped Flightcontrollers that can deal very good with wrong compass heading or complete loss of compass.
So while that messes up your flight its no reason to go haywire at all unless the Software is flawed.
Simple check: Look at the heading of your craft on the map.
Next check: hover and the give little stick inputs forward / back / left right.
It should halt very quickliy and not start circling (aka toilet bowl)
Last check, after you took off, just yaw while otherwise hovering in > 13m Altitude (so the VPS is ineffective).
It should stay in place, not do circles.
If thats okay your compass is okay, otherwise land ASAP
And to just rub it in again, a $50 APM can cope with bad / lost compass and still do a half decent RTH.
And by the way, Alt hold ( == ATTI) is completely okay even then.
Still the question is: Does the Mavic "think" there is stick input or is there a "ghost" input from the controller.
If the latter is true switching off the Tx would enter a clean RTH. If that doesnt help its the mavics Bug.
Jason,
Was the crash your last flight?
I'm looking at the last flight and the only thing I see is a climb than motor obstructed.
It shows a takeoff without GPS lock, my device cannot get all the data.
Let's wait and see what the flight data analysis team says after they analyze your DAT file.
Too bad there is no HEADING INDICATOR on this AIRCRAFT. That would be superb for several reasons along the top of the screen.
I believe iOS has a super basic one with a poor refresh rate. That weird "radar" thing on iOS with the hilarious BLUE for the ground/bottom portion.
Isnt there a heading in the little maos view ? Stil beeing a Mavic virgin i cannot try but saw it on many screenshots. No way precise but a first hint.
Only on iOS:
- There's a little 'radar' map thing that only has an "N" for North on it. It's too small and doesn't refresh fast enough to be useful for getting an actual feeling of your faster-than-slow heading changes.
- There is an inaccurate attempt at an artificial horizon inside this 'radar' map thing (lol). It shows BLUE for the GROUND - I'm not sure if that's hilarious or scary.
I mean, it's like there aren't thousands of aviation experts, beyond that of flying R/C aircraft, available to hire as maybe just a part-time consultant. Or Chinese-to-English translators for that matter.
DJI-Ken Posted at 2016-11-7 04:22
Jason,
Was the crash your last flight?
I'm looking at the last flight and the only thing I see is ...
Yes, I believe you were looking at the correct flight. The app was telling me there was a weak GPS signal when I launched. It said to fly with caution. I felt it would be ok since I was really only going to do a quick flight nearby, and I had previously done several flights indoors, with no GPS signal at all. I thought I should be ok with controlling the drone manually if necessary, as I would have my eyes on it at all times.
I've updated my main post with video of the crash.
Your keen, you took off with weak GPS signal on a night time flight, bad enough in the daytime. Hope it all works out for you with DJI. Be interested on the motor obstruction, do you have bats around the area, that type of noise from the Mavic will set them off flying?
Are these drones known to cause issues with bats? We do see them around here, but I don't think there's a colony or anything like that nearby.
Yes, I suppose I was a bit keen when making this flight. And honestly, if this crash is totally due to my own bad judgement, I'll accept that. But when I started hearing about other similar issues, especially with the left stick, I thought I should post my experience, in case it is due to a larger problem.
mate not to worsen your situation but that was one risky flight, even with GPS, a night time flight can be tricky (not to mention illegal as per CASA unless you have the license) and without GPS, it is downright dangerous. Hopefully you had DJI Care Refresh?
a point on insurance after reading this..
I just contacted my insurance company here in New Zealand AA insurance that I have my household contents with.
The drone or any remote controlled model is fully covered for accidents or fly away loss.
Full replacement cover of the drone, all i pay is a $300 excess and they will replace it in full. I told them it was worth $2000 and they said np it would be replaced in full.
They pointed out though that there is no public liability cover for any damage to property or persons it causes during the accident incident..
I would be calling your insurance company to find out as it flew away and had an accident outside your control.
better than DJI care refresh as I don't think it is available in NZ anyway..
lvl.4
lvl.1
Second Officer
Captain
DJI team
lvl.3
lvl.2
everything costs more for us so I kinda was expecting that price tag.