autotest: bad fix for failing Sub test #12886
Conversation
There is something significantly wrong with Sub's manual input at the moment. Stepping the input up to 1525 yields sensible results. 1526 is a break-value - spin goes up dramatically. The test currently flaps as it is random as to whether any given VFR_HUD packet will be at the expected heading - this thing spins very, very fast.
|
It looks like the vehicle is not actually spinning very fast. Looking at logs, it is turning ~2 rad per second. It appears that the I think this is the problem, but I'm not sure where to look from here. |
|
Looking further into the log I found this:
|
|
2 radians per second is pretty darn fast for a sub :-) 360 degrees/second, VFR_HUD is at 4Hz, meaning we'll see the heading every 90 degrees or so :-)
That's me stepping up the RC input in 1pwm increments. As soon as it gets out of the DZ it starts to spin at 50 degrees/second. Is that expected? Did we sort out what was going on with speedup? |
|
Closed in favour of the referenced PR. |
There is something significantly wrong with Sub's manual input at the
moment.
Stepping the input up to 1525 yields sensible results. 1526 is a
break-value - spin goes up dramatically.
The test currently flaps as it is random as to whether any given VFR_HUD
packet will be at the expected heading - this thing spins very, very
fast.
@jaxxzer - we need to apply this very, very bad fix, or perhaps disable the test; CI randomly fails for people ATM.
It's a pretty serious regression that the test seems to have caught, however.