This resolves the issue of the parachute servo's position not being
initialised to the off position due to an ordering problem of the
auxiliary servos being initialised after the parachute object.
AP_OpticalFlow.h becomes simply a file that includes all other optical
flow header files.
OpticalFlow class simplified to only return surface quality, raw output
and velocity vector.
The DCM drift correction code uses the current attitude to calculate
error values to update its gyro drift correction. If we were using EKF
then without this patch the DCM code running as an alternative AHRS
source would be using the EKF attitude for calculating the error
value, leading to very bad gyro drift estimation
use_compass() and reset() are common to AP_AHRS_DCM and
AP_AHRS_NavEKF. As AP_AHRS_NavEKF is a child of AP_AHRS_DCM, when we
call use_compass() from within AP_AHRS_DCM we actually end up calling
AP_AHRS_NavEKF::use_compass().
This has the effect of disabling the compass in DCM when EKF is active
and EKF has decided not to use the compass. That means that the DCM
yaw (and in fact the whole attitude) can get badly off while EKF is
enabled, making DCM an ineffective fallback if EKF fails.
The fix is to call the specific class versions of use_compass() and
reset()
This replaces the ardupilot only NAV_GUIDED command.
Also remove support for NAV_VELOCITY mission command which will be
replaced by SET_POSITION_TARGET non-mission command.
this prevents a floating point error caused by using an uninitialised
vector3 when switching between DCM and EKF control in AP_InertialNav
Pair-Programmed-With: Randy Mackay <rmackay9@yahoo.com>
limit.throttle_lower flag becomes true when the throttle passed into the
motors lib (which is in the 0 ~ 1000 range) is below _min throttle.
This makes the interpretation of the THR_MIN parameter consistent
between the main code (which uses 0 ~ 1000 range) and the motors lib
(which previously used the RC3_MIN ~ RC3_MAX range).
The remaining problem however is that the output of the motors continues
to use THR_MIN as if it were a pwm. I don't believe this is a dangerous
problem however.
The GPS glitch offset was being zeroed during position resets. This caused the filter to reject subsequent GPS measurements if the GPS error persisted long enough to invoke a timeout and a position reset.