Loose of one landing gear led to a crash

[Quadbee]

I´ve lost one of four landings gears (3d printed add-ons) during an indoor testflight and the consequence has been that the drone has started climbing and couldn´t be stopped via remote control until it crashed with the ceiling. The weight of one landing gear is about 5g.

Now I´m trying to find an explanation for this behaviour. Can it be that the IMU measures the needed thrust for hovering after start and when the weight changes during flight that it gets a problem which leads to a destabilization?

 

[ekky-thump]

You will almost certainly be asked for the flight logs, txt and DAT. There is an explanation of their retrival on Mavic Flight Log Retrieval and Analysis Guide

 

[Quadbee]

Thanks for the hint. I´ve attached the log.

The floor was very even in color and shape and it could be that the downward cam couldn´t get a good reference picture.

 

[DanMan32]

It makes sense that the weight imbalance may have confused the IMU.

 

[lannes]

It's probably a combination of magnetic interference and VPS resolution from being too close to the ceiling in a dimly lit room.

If you have a second storey where you fly, it probably has rebar in the concrete ceiling/floor above.

 

[Quadbee]

I´ve flown a couple of times in this room without any problems but the only thing this time was different was that I flew over a red bedsheet. There the Mini lost the landing gear and started to loose control. Maybe the infrared light from the VPS couldn´t be seen by the cam due to the red underground. In this case the only way for the drone to keep altitude is maintaining the same thrust as before and when then suddenly the weight is changing this could lead to a problem. I don´t know how precise the barometric sensor is but I don`t think that it can measure indoor a difference between 1-2 meter.

 

[PhantomFandom]

Now I´m trying to find an explanation for this behaviour. Can it be that the IMU measures the needed thrust for hovering after start and when the weight changes during flight that it gets a problem which leads to a destabilization?

No, there is no such measurement at launch. The IMU dynamically calculates its orientation live as the craft is flying. It depends on the gyro and accelerometer to maintain a stable flight attitude at all times. You could (if practical) load or unload weight during flight and it would compensate (within the effective thrust range of the motors).

In this case the only way for the drone to keep altitude is maintaining the same thrust as before and when then suddenly the weight is changing this could lead to a problem.

No, the IMU will not "maintain the same thrust as before" because we fly in a dynamic 3-D environment. The FC must constantly monitor all sensors (including IMU) to arrive at the proper thrust solution for each ESC. If you had a way to lock each motor at a constant thrust (RPM), I am sure the drone would soon meet its demise, in any conditions.

I don´t know how precise the barometric sensor is but I don`t think that it can measure indoor a difference between 1-2 meter.

Yes it certainly can.

 

[aeropic]

5g of unbalance is very few... That's weird!

 

[Zbip57]

Did the landing gear fall clean off, or was it left dangling in sight of the VPS sensor?

If you have the Mini hovering in front of you and stick your hand underneath it, the VPS sensor detects that it's suddenly too close to the "ground" (your hand) and will climb all by itself to a "safe" height. If the detached landing gear was hanging in view of the VPS sensor, maybe the Mini was climbing trying to clear that obstacle.

 

[Deleted member 103366]

Ever being one for daft experiments, I attached 20 x 1" paper-clips to a piece of thread attached to one front landing leg. The thread was 1m long so the mini was well-airborne before it had to lift them. They made not the slightest difference to the attitiude of hover etc. I even repeatedly lifted the paperclips suddenly and dropped them, lifting did nothing, dropping them caused the Mini to twitch but recovery was 'instantaneous'

 

[diver dave]

I dont think the loss of leg weight was the issue. I tried a drop kit on my mavic pro for the first time last week, a small bag of gravel, a target and a stupid test for a laugh. It flew perfectly, though it was clearly noisier as it needed more thrust to lift the bag. First test was at about 20 foot, dropped the bag and it shuddered and stabilised, no change. Next test was far higher, probably 100 foot, some wind, and same result, so with this massive weight change the difference was basically a shudder and nothing else, except two 50 plus year olds giggling like kids!
I expect something else was your issue, as the difference in % weight would have been far greatr in ur test tham a MM losing a 5g leg, i have lost legs from my MM and MP in flight and not noticed until it landed, and tipped. I am finding masses of issues every time i fly indoors.

 

[hiflyer201]

The only non commanded and un controlled climb that I've had was because I was careless putting on a skin and very slightly covered the edge of a sensor on the bottom of my Spark. It took off and shot up like a rocket..Dirt on the sensor will cause trouble also

 

[Quadbee]

Thanks for all the answers!

It can be that it doesn´t have to do anything with the loss of weight but then it was in my opinion caused by a problem with the VPS. As described I´ve been flying over a red bedsheet when the bird suddenly lost the landing gear and started to rise. Perhaps the infrared spot couldn´t be seen by the cam in this case and the change of weight couldn´t be compensated. I checked the cam afterwards but couldn´t see any obstacles in it. Maybe it was dust or a small grain of sand.

What I really don´t like when the commands by the pilot could be overruled by the flight computer which was the case at my crash. I´ve had whole the time the pitch stick down when the strange behaviour began but some wrong interpretation of the IMU signals has been stronger than the human input. Thats definetely not good and should be changed by DJI otherwise those devices are more dangerous than believed.