Crashed , bottom sensors to blame

[msmst25]

IMG_0691

• msmst25
• Jan 1, 2017
• 1

Here is the result of my attempt to fly indoors in ATTI mode. It suddenly started flying backwards and wouldn't respond to land command and I tried to stop it with my hand to keep it from running into anything.

 

[f3honda4me]

I'm not sure if that was to me specifically or the OP; however, I figured more data never hurts... This is from the iPhone app not the Mavic. Does that really matter? Strange that the DJI support wanted me to directly upload the Go4 logs (had already downloaded manually and removed the app from that device) and did not suggest the Mavic directly...

You can see it is jumping in and out of ATTI mode..

ExpireBox | FLY017.DAT

I also uploaded the logs to Phantom Logs - Phantom Log Viewer - PhantomHelp.com

Hmm this DAT file says unable to determine aircraft type. The DAT file is from the mavic itself, correct? Did you follow these steps to get it? How to retrieve a .DAT

Are you on a mac or pc?

 

[MavicPilot]

Hmm this DAT file says unable to determine aircraft type. The DAT file is from the mavic itself, correct? Did you follow these steps to get it? How to retrieve a .DAT

Are you on a mac or pc?

I do apologize for any time wasted. As mentioned, this was downloaded from iTunes via the GO 4 logs. I just downloaded the Mavic log file as per link (Thank you!) - first time I have ever done this/needed to do this.... My P3A was always updated OTA and logs OTA.

ExpireBox | DJI_ASSISTANT_EXPORT_FILE_2017-01-01_21-17-27.DAT

The file is named as the export date, but the event was on 12/12/16.

 

[MavicPilot]

IMG_0691

• msmst25
• Jan 1, 2017
• 1

Here is the result of my attempt to fly indoors in ATTI mode. It suddenly started flying backwards and wouldn't respond to land command and I tried to stop it with my hand to keep it from running into anything.

Oh wow! Dog (larger breed) got to close to the blade by jumping up to investigate (high jumper...) and got hit at a regular RPM speed. I saw it coming mid jump and thought throwing it up (throttle) would increase the speed and cause more damage. No yelping, blood or injury. I've gotten hit by my original Phantom a couple of time without issue, but thank you for sharing and it looks a couple of days old.

Tip - if you wrap a wound early and keep it wet (petroleum jelly, neosporon, etc, it will not scar up and heal a lot better - that looks like a couple of days old and scabbing over). Wet means pretty much always so 3-4x a day changing the bandages.

 

[f3honda4me]

Looking at the dat file, being perfectly honest, it kinda looks like it started to drift in ATTI mode, and you paniced and hit the sticks in a bad way. At 83.2 seconds in the log it went into ATTI. This was during that last yaw left we see on your video. At 86.78 seconds the craft records an in air impact. Between the 83.2 second mark and 86.78 mark, you push the elevator (right stick) all the way down and then up, as well as some aileron movements. At 85.5 seconds you start pushing the right stick forward. From 86.08 to the moment of impact you are holding the right stick forward all the way which sent the drone into the wall/tv.



See screenshot plus log event of impact

86.783 : 505162092 : 5393 [L-FDI][CTRL]: fault on , impact_in_air

 

[MavicPilot]

Looking at the dat file, being perfectly honest, it kinda looks like it started to drift in ATTI mode, and you paniced and hit the sticks in a bad way. At 83.2 seconds in the log it went into ATTI. This was during that last yaw left we see on your video. At 86.78 seconds the craft records an in air impact. Between the 83.2 second mark and 86.78 mark, you push the elevator (right stick) all the way down and then up, as well as some aileron movements. At 85.5 seconds you start pushing the right stick forward. From 86.08 to the moment of impact you are holding the right stick forward all the way which sent the drone into the wall/tv.



See screenshot plus log event of impact

86.783 : 505162092 : 5393 [L-FDI][CTRL]: fault on , impact_in_air

BTW, I was not the original OP/poster. Thank you for taking a look!

I had a similar scenario and I think there is a serious software/hardware glitch with inside flight and specifically around the craft going in and out of ATTI/GPS mode. I wanted to upload my version to compare it to the OPs or to add it to others that may already exist.

Assuming you are looking at my logs and wanted to see the video - I can edit the video and post it. It has a lot of family photos that I will need to edit out for anonymity, but I do have it. I can tell you that...

1. It was drifting in ATTI mode the entire time.
2. I may have paniced a bit, but only because the RC was not transmitting data (my perception based on events). For a split section I did hit it a bit harder left (mode2) than it needed as I was 2-3' from the wall.
3. The RC/Mavic did NOT accept/respond the command and this led to me pushing into stick more than I would have normally (as per #2 no more). I admit that part, but only due to lack of a response
4. I let go of all buttons/sticks/etc on the RC (as we would do in any case we are not happy or paniced (or at least what we should do if paniced)). The craft was still hovering/drifting towards the wall (rear of craft I believe).
5. 1-2s later the craft banked hard (hand no longer on the RC). It seemed like a FULL left motion.


What tool did you use to generate the data? I appreciate the photo, but without some data I am not sure how to read/interpret it.

 

[f3honda4me]

BTW, I was not the original OP/poster. Thank you for taking a look!

I had a similar scenario and I think there is a serious software/hardware glitch with inside flight and specifically around the craft going in and out of ATTI/GPS mode. I wanted to upload my version to compare it to the OPs or to add it to others that may already exist.

Assuming you are looking at my logs and wanted to see the video - I can edit the video and post it. It has a lot of family photos that I will need to edit out for anonymity, but I do have it. I can tell you that...

1. It was drifting in ATTI mode the entire time.
2. I may have paniced a bit, but only because the RC was not transmitting data (my perception based on events). For a split section I did hit it a bit harder left (mode2) than it needed as I was 2-3' from the wall.
3. The RC/Mavic did NOT accept/respond the command and this led to me pushing into stick more than I would have normally (as per #2 no more). I admit that part, but only due to lack of a response
4. I let go of all buttons/sticks/etc on the RC (as we would do in any case we are not happy or paniced (or at least what we should do if paniced)). The craft was still hovering/drifting towards the wall (rear of craft I believe).
5. 1-2s later the craft banked hard (hand no longer on the RC). It seemed like a FULL left motion.

Ah ok you're not OP. But there were definitely inputs by the sticks all the way to moment of impact. I'm still learning to use datcon, so there might be more @BudWalker could tell you but I know there were input commands for sure.

It's normal for the craft to go in and out of gps and atti indoors.

 

[MavicPilot]

Ah ok you're not OP. But there were definitely inputs by the sticks all the way to moment of impact. I'm still learning to use datcon, so there might be more @BudWalker could tell you but I know there were input commands for sure.

It's normal for the craft to go in and out of gps and atti indoors.

Thanks for the app name! I will definitely need to dig into that to analyse future flights.

I am not denying your interpretation of the data, as I am not versed in it, but anodically I can tell you that this was not the case. I have seen a couple of posting here where users have reported the RC not responding, so I wonder if the swap between GPS and ATTI could cause a slight communication overload/delay?

In my case, and I will never trust indoor flying again, it seems like some program/command (GPS/ATTI or ATTI/GPS transition) was running and taking the processor priority and it had queued up the other commands (user input in this case...). Once it (? no idea what for sure) finished, it ran through the queued up commands. In this case, because it was delayed it seemed like it interpreted "too much"/"more" information on the direction I had directed it to.

For anyone that has use RDP/MSTSC/VNC/PuTTY/ETC to control a remote PC. You may have see that sometime the keyboard sends extra characters - as if the user sent them. Yet, the user only sent the keystroke a single time. This has happened frequently enough to me over the past 10 years, but for someone that only uses it once or twice it may never be seen.

IMHO, we are seeing the same behavior/latency here. The single H (for http as an example) that gets sent to the remote computer and interpreted as HHHHHHHHHHHHHHH (even with a single key stroke on the local PC) is a prime example of what I feel is happening on the Mavic. So, the user/RC is sending a small % (however it is calculated) of Left; however, due to the delay (or even without) it is being multiplying out to some much greater degree (10% vs 70% max as an example).

I hope that makes sense, but it was the only method that I could try comparing it to other scenarios that I have seen the same results. The Mavic runs on software and communicates over wifi / protocols / packets, so this also is a pretty sound comparison.

Thanks again!

 

[Tok Pilot]

I have flown mine indoors many times, even in a tight space with many people sitting close by in a 4 foot square, even grabbed one of the landing legs and drug it a foot away just to watch it go right back to where it was hovering and never had a problem, surprisingly it actually flies more stable indoors than outdoors if you can believe that, doesn't drift at all.


Sent from my iPhone using MavicPilots

 

[MavicPilot]

I have flown mine indoors many times, even in a tight space with many people sitting close by in a 4 foot square, even grabbed one of the landing legs and drug it a foot away just to watch it go right back to where it was hovering and never had a problem, surprisingly it actually flies more stable indoors than outdoors if you can believe that, doesn't drift at all.


Sent from my iPhone using MavicPilots

Did yours go in and out of GPS/ATTI mode? In my case, I was in a room with an entire wall (23' long) with 4' long windows (~2" between and likely 7-8" glass-to-glass including trim). I am not doubting that DJI did some indoor testing, but advocating that there are issues that seem common and require addressing.

 

[Tok Pilot]

No, I wasn't in a room with windows enabling the MP to pick up and lose satellites, stayed in ATTI mode the whole time, actually went into tripod mode once while flying on doors, but haven't had that option but once.


Sent from my iPad using MavicPilots

 

[bakerboy]

If you want to fly indoors then you better know how to fly without sensors, go get a toy quad and practice

 

[Chase R]

Indoor flying requires a lot of light. Even dim rooms will mess with the sensors and it becomes unstable.

 

[MavicPilot]

If you want to fly indoors then you better know how to fly without sensors, go get a toy quad and practice

I am not the OP/poster, but I agree with you wholeheartedly. My complaint is the RC loosing (or delaying) the commands sent to the quad due to potentially it swapping in and out of gps/atti mode. I will leave it there as I already wrote up a huge piece on my feels on the scenario, but I have no way to prove it.

For reference, I have a two parrot quads, a couple of the micro ones(CX-10 and the like), and two heli's that I have been flying inside for almost 10 years now (blade cx2 and a Syma S107). I also have the original phantom (flown indoors but only hovered due to the size) and a P3A that I never even tried indoors. Both of those are used in ATTI mode for certain shots and the original Phantom was nowhere near as stable as the P3A or the Mavic.

 

[J-Dub]

Anybody know what settings they use to fly indoors for exhibition expos??

 

[bakerboy]

I am not the OP/poster, but I agree with you wholeheartedly. My complaint is the RC loosing (or delaying) the commands sent to the quad due to potentially it swapping in and out of gps/atti mode. I will leave it there as I already wrote up a huge piece on my feels on the scenario, but I have no way to prove it.

For reference, I have a two parrot quads, a couple of the micro ones(CX-10 and the like), and two heli's that I have been flying inside for almost 10 years now (blade cx2 and a Syma S107). I also have the original phantom (flown indoors but only hovered due to the size) and a P3A that I never even tried indoors. Both of those are used in ATTI mode for certain shots and the original Phantom was nowhere near as stable as the P3A or the Mavic.

Piece of copper tape of the gps will make sure it doesn't gps signal inside

 

[Tok Pilot]

I don't use any "settings" while flying indoors, the MP knows it's indoors if it isn't picking up any satellites, goes into ATTI or sometimes "OPTI" mode and uses the sensors to position itself where it is when it turn loose of the controls, it's very impressive that way.


Sent from my iPhone using MavicPilots

 

[J-Dub]

I don't use any "settings" while flying indoors, the MP knows it's indoors if it isn't picking up any satellites, goes into ATTI or sometimes "OPTI" mode and uses the sensors to position itself where it is when it turn loose of the controls, it's very impressive that way.


Sent from my iPhone using MavicPilots

However, for many it does get GPS indoors (I regularly get 5-8 Sats inside, picks them up thru the windows) and then all is great until the MP loses enough GPS-lock to automatically revert to ATTI, causing unwanted drifting and crashing if the pilot cannot control it fast enough. If DJI would make a switch to Not Attempt To Locate Satellites, then indoors would not be an issue.

 

[Tok Pilot]

If the GPS on mine was working properly, it would probably do the same thing , but unfortunately I think mine is defective


Sent from my iPhone using MavicPilots

 

[erikgraham]

Wait, you guys are actually flying indoors - as in inside a normal house in normal rooms? Seems kind of crazy. I'm new around here and new to the Mavic, but I've flown indoors with optical flow on DIY built drones many many times. But that was in warehouses, sports complexes, and other large spaces. It would never occur to me to fly in my house, other than the garage for a quick test.