I had a horrible indoor crash today after I armed my motors.
I was in stabilize mode, the quad changed direction towards one side and crashed into the wall, broke 2 props, 2 motor shaft bent. I have to check the 20Amp ESC's if they got burnt in the process due to over current :( Bled my arm in order to save the other two props!!
I had my GEO Fence ON. After I see the log, it shows has triggered a fail-safe (RTL) as I breached the fence? I am unable to interpret why this has happened? Outside, it flew like a charm the last time I flew.
Why did it breach the fence. It was supposed to be in Stabilize. No waypoints, simple indoor flight in STAB mode.
Subsystems/ Error codes:
9: Fence, ECode 2: circular fence breached
Log attached, any comments will be helpful
I'm not an expert but from the log it looks like even though indoors your quad was still using GPS although with a really bad hdop. Its position was bouncing around. If you plot the lat/long coordinates it moved quite a bit. It simply thought it moved beyond the circular fence you had enabled, and unfortunately it was armed.
If this is the case it highlights a danger with the fence enabled and poor GPS coverage. Maybe there should be a condition of a minimum hdop or number of satellites before the circular fence failsafe kicks in? And at the very least if doing indoor testing disable the fence failsafe!
You are right, and yes, this is unfortunate. IMHO, the software needs to address this.
I had plugged in the battery in my lab (some 30m away from where I armed it).
I armed the motor and gave it throttle (~50%) and it swung immd to one direction and I COULD NOT stop it using Txm roll/ pitch sticks. It just crashed!
Few important questions:
1) Is the GEO Fence location set from the place where it is armed or from the place where we plug in the LIPO (i.e. switch on) the APM?
2) Any idea from the logs on 'why' the Geo fence was triggered in the first place. These are the log params:
PARM, FENCE_ENABLE, 1.000000
PARM, FENCE_TYPE, 3.000000
PARM, FENCE_ACTION, 1.000000
PARM, FENCE_ALT_MAX, 20.000000
PARM, FENCE_RADIUS, 100.000000
And I was, of course well within it. I just came out of the lab to fly it.
Shyam, the fence location and the "home" location are set from where you arm. But indoors that could be just about anywhere, and it may change radically from minute to minute. You do not want any GPS-using mode or feature running indoors! (Whether anything at all should be run indoors is another questions...) This is a fundamental mistake and I don't think needs any software fix. Also, practice, practice, practice switching into manual mode instantly the second anything goes weird. I've slowed down the travel and response rates to give more time to observe behavior before hitting the panic button. Once I'm really comfortable I'll maybe speed things back up.
When you say "it can change radically from minute to minute", why would this happen as 'arming' is under the user's control and when it gets the GPS lock, it should set the FENCE location, am I correct? I understand, there are reflections inside, in my case I had 6 satellites with 3D lock constant. What happens when I lose the GPS lock, it should trigger an RTL, am I correct?
I am comfortable flying outside, and gives me sufficient time before hitting the panic button, but indoors the reaction time reduces to less than a second because copter does not have enough height anyways before a potent crash!
This brings the question, when we have Optical flow sensor, and we test indoors, we disable the GPS?
Shyam, I think the lesson is not to use ANY mode that uses GPS indoors. Even with GPS lock, your GPS will report bad positioning because of the low number of SATs it can receive. Enough for GPS lock, but not enough to be accurate.
I ONLY use stabilise mode indoors.
I did not use any mode using GPS. Rather, I used only Stabilize mode since I was flying indoors. But, this crash is kind of a mystery that I am unable to resolve.
The question I am wondering is "Why did it break the GEO Fence indoors in the first place". Ideally, it must not.
And yes, to add to that:
Just before I tried to stop everything, I switched the throttle down to 0. I use 9-CH FlySky Tx/Rx. The props did not stop. This can be seen from the log. Any comments on this?
The props (throttle) are not under the control of your Tx once the aricraft has switched itself into a full autonomous mode such as RTL (here by hitting what it thought was the fence). You need to switch out of stabilize and then back into stabilize. That would have returned control to you. That's why I said earlier that getting that move down so it can be executed quickly, without having to think about it or fumble around, is important.
Shyam, if you have the fence enabled, you are then flying in a GPS-dependant regime, regardless of what your "flight mode" is. Being indoors with reflections of the GPS signals etc. apparently resulted in APM concluding that it had moved outside of the fence. As soon as that happens whatever is set up to occur at fence-breech takes over, for example RTL. So, you don't want the fence enabled while flying indoors, or if it is enabled you want to only get a report, not an action.
Not adding any tech advice, but just like Hitchhicker's when flying don't forget your towel. Chuck it over the flailing beast to contain it. Perhaps there should be 3DR towels.
How would GPS ever be accurate enough for the confines of a small room?