Ran into a bug on our physical plane where failsafe.last_valid_rc_ms was not recognizing that the transmitter had failed. This is likely due to how the standard failsafe works in receiving lower-than-possible throttle values. So in order to account for this, I added a new variable to the failsafe, AFS_last_valid_rc_ms, and I update it only if the ch3_failsafe (the throttle failsafe) is not on. If the throttle failsafe is on, that means that the plane has indeed lost transmitter input, so the AFS needs to recognize that.
/home/travis/build/ArduPilot/ardupilot/libraries/AP_Param/AP_Param.h:542:22: warning: using floating point absolute value function 'fabsf' when argument is of integer type [-Wabsolute-value]
bool force = fabsf(_value - v) < FLT_EPSILON;
@Description: This parameter reduces the pitch minimum limit of an auto-takeoff just a few seconds before it reaches the target altitude. This reduces overshoot by allowing the flight controller to start leveling off a few seconds before reaching the target height. When set to zero, the mission pitch min is enforced all the way to and through the target altitude, otherwise the pitch min slowly reduces to zero in the final segment. This is the pitch_min, not the demand. The flight controller should still be commanding to gain altitude to finish the takeoff but with this param it is not forcing it higher than it wants to be. (+1 squashed commits)
threading bug fix. When a mission wp updates, but the L1 controller had not yet, the data is stale. Example, On Plane when NAV_LAND starts for a moment your xtrack and bearing is most likely bear zero regardless if you have a big turn or not until 10 Hz later when the update() gets called and updates those values with correct values for the new waypoint.
recent fixes in Plane have made the stage more accurate so exceptions/hacks are no longer needed to differentiate between knowing if executing NAV_LAND vs being in stage_approach.
the glide_slope gets calculated every time there's a major event such as mission item change or wp_proportion change so its good to update the flight stage then too because.
also logging stage when stage changes, might as well get an extra data point in there when it's timely
Luis Vale Gonçalves
Randy Mackay
Andrew Tridgell
Tom Pittenger
James Stoyell
Francisco Ferreira