Running Arducopter 3.0 on an APM 2.5, 3dr Quad.
Testing out Geofence altitude limit. Stabilize mode, climb to Geofence alt (set for 50). Geofence engages RTL, quad starts to land. Just as the quad landed (just touched down, motors still spinning a little) the quad all the sudden throttled way up and pitched over sideways, flipping upside down. I had to switch flight modes from Stabilize to AUTO and back to stabilize before it would let me disengage the motors (they were pinned to the ground letting out a high pitched whine from the props being stopped).
I looked back over my logs and it looks like the battery voltage failsafe was engaged. I have had this happen before where I was just about to land (inches away from the ground) only to have the battery failsafe engage and the quad try to climb to my RTL altitude.
Maybe give the failsafe an escape where it says "If I'm already doing an AUTO type landing (RTL or LAND), then just continue the landing sequence if a battery failsafe is engaged."
Event is at about t=900 on the logs, you can tell where it is by the sudden flipping of the Z-acc from the copter flipping upside down.
Also, unrelated problem, but sometimes when I am flying an AUTO mission the quad points just a little off of the direction of travel. Second .tlog (2013-07-02 13-13-40.tlog) has a AUTO mission where it does this. This is just sometimes and will come and go throughout a mission. WP_YAW_BEHAVIOR = 1.
And another question. The beginning of my .log files used to look very nice and neat, now it looks like a bunch of random junk. All of the data is fine, but the first few lines above the parameters appear to be corrupt.
Replies
Logan,
Good point about it not kicking off RTL again if you have a battery safe when you're already in RTL. It's a very simple fix so we will put that in for 3.1.
Needing to change the flight mode switch to another mode and then back to stabilize is pretty much standard procedure in this situation. We only allow you to disarm in stabilize or acro which makes sense because you can imagine situations where the user could input zero throttle and left rudder in other modes like loiter without meaning to disarm the motors.
The issue with the copter pointing in slightly the wrong direction is because it heads in the direction that the next waypoint was at the moment the copter completed the previous waypoint. So if the copter is swept downwind a bit it can end up facing in slightly the wrong direction. The reason it's done this way is we didn't want the copter to waste energy always turning to face the next waypoint. For the most part it doesn't really matter which way the copter faces anyway.
The junk above the parameters is required to allow the mission planner to use the correct column headers always. Previously there was always an issue when we updated the dataflash log format that the mission planner's column headers would mismatch either the old version or the new version.
Thanks for the report. I sounds like the safe route is to set the low battery function to land in stead of RTL until this is resolved.
I have also considered raising the low voltage level to allow for enough current for an RTL.
I will check all my settings again.