AP_AHRS_NavEKF doesn't use the rangefinder, it just delegates the
calls to the right estimator.
For libraries/AP_AHRS/examples/AHRS_Test/AHRS_Test.cpp the initialization
order has also been modified to match the order on vehicles. It's more
correct since it passes a reference when EKF2/EKF3 are already
constructed, while thos constructors use a pointer. Ideally these
should be moved to an init() method though, or use a get_instance().
We should never include version.h or ap_version.h headers directly
on a header since this will trigger a complete rebuild of the
codebase when we commit to the repository. The ap_version.h header
is auto-generated containing information from the current commit.
If we include it in a header, every other file that ends up including
that header (directly or indirectly) will need to be rebuilt. No
ccache's cache beats having to do nothing when the header is just
not included.
version.h contains information that is kept on a struct inside
each vehicle. Rather than using the macros from each vehicle,
the getter should be preferred, which returns an AP_FWVersion
referente.
this fixes the OVERRIDE_CHAN functionality to work correctly with the
new function based mixing code
It isn't a perfect match for ArduPilot mixing, but it is very close
for the key control surfaces, including mixed surfaces
the high loop rate of quadplanes led to less than 1cm/loop change in
height, which got truncated to zero. Adjusting height at 10Hz fixes
that.
Thanks to Marco for reporting this!
some people connect USB to allow connections from a companion
computer. The arming check is sufficient to prevent unwanted battery
failsafes when bench testing
the nav controller can think we have already reached the loiter target
if we were last in a LOITER when we switch to RTL. In that case it
would switch to QRTL immediately
found by Leonard (thanks!)
Removes the need for plane to do the math for finding the relative height.
Also caches the value at the same time we update current_loc, which is a
non behaviour change as that was the only time you could see a change in
the relative height propegate through the system anyways
- move functions restart_landing_sequence() and jump_to_landing_sequence() to AP_Landing
- NOTE: jump_to function can not set mode, so it is now done externally in vehicle
This doesn't count repeated readings towards the rangefinder count,
and resets the counter if the change is more than 20% of full
range. This greatly reduces the impact of poor sonar sensors for
landing
Moved the code theat updates the home position while the aircraft is
unarmed from 10hz loop to one second loop and ensured that is does not
update more then once every 5 seconds. Closes issue 4311.
new param: LAND_ABORT_DEG
@Description: This parameter is used when using a rangefinder during landing for altitude correction from baro drift (RNGFND_LANDING=1) and the altitude correction indicates your altitude is higher than the intended slope path. Steeper slopes can result in crashes so this allows the option to remember the baro offset and self-abort the landing and come around for a another landing with the correct baro offset applied for a perfect slope. An auto-abort go-around will only happen once, next attempt will not auto-abort again. This operation happens entirely automatically in AUTO mode. This value is the delta degrees threshold to trigger the go-around. Example: if set to 5 deg and the mission planned slope is 15 deg then if the new slope is 21 then it will go-around. Set to 0 to disable. Requires LAND_SLOPE_RCALC > 0.
New param: LAND_SLOPE_RCALC
@Description: This parameter is used when using a rangefinder during landing for altitude correction from baro drift (RNGFND_LANDING=1) and the altitude correction indicates your altitude is lower than the intended slope path. This value is the threshold of the correction to re-calculate the landing approach slope. Set to zero to keep the original slope all the way down and any detected baro drift will be corrected by pitching/throttling up to snap back to resume the original slope path. Otherwise, when a rangefinder altitude correction exceeds this threshold it will trigger a slope re-calculate to give a shallower slope. This also smoothes out the approach when flying over objects such as trees. Recommend a value of 2m.
default value is 2 (so, enabled by default)
Further to refactor of RC_Channel class which included
adding get_xx set_xx methods, change reads and writes to the public members
to calls to get and set functionsss
old public member(int16_t) get function -> int16_t set function (int16_t)
(expression where c is an object of type RC_Channel)
c.radio_in c.get_radio_in() c.set_radio_in(v)
c.control_in c.get_control_in() c.set_control_in(v)
c.servo_out c.get_servo_out() c.set_servo_out(v)
c.pwm_out c.get_pwm_out() // use existing
c.radio_out c.get_radio_out() c.set_radio_out(v)
c.radio_max c.get_radio_max() c.set_radio_max(v)
c.radio_min c.get_radio_min() c.set_radio_min(v)
c.radio_trim c.get_radio_trim() c.set_radio_trim(v);
c.min_max_configured() // return true if min and max are configured
Because data members of RC_Channels are now private and so cannot be written directly
some overloads are provided in the Plane classes to provide the old functionality
new overload Plane::stick_mix_channel(RC_Channel *channel)
which forwards to the previously existing
void stick_mix_channel(RC_Channel *channel, int16_t &servo_out);
new overload Plane::channel_output_mixer(Rc_Channel* , RC_Channel*)const
which forwards to
(uint8_t mixing_type, int16_t & chan1, int16_t & chan2)const;
Rename functions
RC_Channel_aux::set_radio_trim(Aux_servo_function_t function)
to RC_Channel_aux::set_trim_to_radio_in_for(Aux_servo_function_t function)
RC_Channel_aux::set_servo_out(Aux_servo_function_t function, int16_t value)
to RC_Channel_aux::set_servo_out_for(Aux_servo_function_t function, int16_t value)
Rationale:
RC_Channel is a complicated class, which combines
several functionalities dealing with stick inputs
in pwm and logical units, logical and actual actuator
outputs, unit conversion etc, etc
The intent of this PR is to clarify existing use of
the class. At the basic level it should now be possible
to grep all places where private variable is set by
searching for the set_xx function.
(The wider purpose is to provide a more generic and
logically simpler method of output mixing. This is a small step)
Having the version macro in the config.h and consequently in the main
vehicle header means that whenever the version changes we need to
compiler the whole vehicle again. This would not be so bad if we weren't
also appending the git hash in the version. In this case, whenever we
commit to the repository we would need to recompile everything.
Move to a separate header that is include only by its users. Then
instead of compiling everything we will compile just a few files.
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.
@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)
- fixes bug where a bungee launch is configured but the aircraft gets bumped and triggers the prop to spin up. This will now detect that and "crash' and disable the motor
Reverse thrust for controlled landings, even with much steeper approach slopes. This is achieved by allowing throttle demand to go negative to maintain a target airspeed. A Pre-Flare stage was added, triggered by an altitude, to allow for a slower airspeed just before land. That lower airspeed can be near stall.
new params LAND_PF_ALT, LAND_PF_SEC, LAND_PF_ARSPD, USE_REV_THRUST
Added update_trigger and check_digital_pin functions
added camera trigger precise time mark
detect camera feedback pin status
added support for simple digital pin
included support for digital pin. Already included in
added support for TRIGGER MSG
@Description: X-Axis deceleration threshold to notify the crash detector that there was a possible impact which helps disarm the motor quickly after a crash. This value should be much higher than normal negative x-axis forces during normal flight, check flight log files to determine the average IMU.x values for your aircraft and motor type. Higher value means less sensative (triggers on higher impact). For electric planes that don't vibrate much during fight a value of 25 is good (that's about 2.5G). For petrol/nitro planes you'll want a higher value. Set to 0 to disable the collision detector.
- inhibit crash detection warnings when disabled by param so now it can be completely disabled
- reset is_crashed when disabled by param
- fixed pre-takeoff detection bug by adding in_preLaunch_flight_stage() where we are actually in FLIGHT_NORMAL instead of FLIGHT_TAKEOFF during setup of bungee launches. This now detects if we're in that state
- simplified the use of been_auto_flying to check across all flight stages. before it was excluded to handle hand-launches which can now be detected with in_preLaunch_flight_stage()
- added impact detector timer to clamp is_flying a few seconds after an impact
- logging new impact detector as "STAT.Hit"