Prior commit added opening of /dev/null as 0, 1 and/or 2 file
descriptors, if they where not present. However, if the temporary
file descriptor used to open /dev/null matched the target file
descriptor, it would be immediately closed again. This commit fixes that,
and does not duplicate and close the temporary file descriptor if it is
already at the correct number.
- new FIFO_RESET state used to give the sensor more time after successful configuration before sampling begins
Co-authored-by: Mathieu Bresciani <brescianimathieu@gmail.com>
* rtl: remove unconditional transition to land after descent
This was a bug, as it renders the above code lines useless.
This would cause a undesired FW landing for VTOL vehicles if
RTL_LAND_DELAY is above 0.
* rtl: head to center after loiter in VTOL FW
To get the same behavior for RTL with and without loiter before land for
VTOL drones.
* rtl: always go to descend state after return
Previously, the state would change directly to land if in MR and
RTL_LAND_DELAY was 0.0, but we will still wish to descent to
RTL_DESCEND_ALT at descent speed, instead of using landing speeds.
* rtl: mark head to center state as part of vtol transition
The next step in the sequence is transition to MC. By setting
vtol_back_transition we ensure that the acceptance radius is adapted to
the expected transition distance.
If 0, 1 and/or 2 file descriptors are not open when mavlink module
starts (as might be the case for USB auto-start), use default /dev/null
so that these numbers are not used by other other files.
Instead of interpret a request for "more logs than currently exists" as
a new request, use a request for index 0, which is more likely to be
the first request.
* [npfg]: Remove the guideToPoint function and replace with guideToPath
* [npfg]: remove unused navigateXXX functions
* [npfg]: Move navigateXXX Function into FWPoscontrol
* [FixedwingPositionControl]: Set default flaps and spoilers in attitude setpoint topic, and only change if necessary.
- precision landing works incorrectly, target position is not updated during the descent above target
- _prepareLandSetpoints needs to update _land_position continuously
Co-authored-by: kapacheuski <kapacheuski@gmail.com>
- the thermal offsets are an optional correction applied to the raw data, so when updating an existing calibration offset with new learned bias we don't want this incorporated
- EKF isTimedOut(), isRecent(), and isNewestSampleRecent() need to handle the case where the timestamp has never been set
- reset() more thoroughly reset fields (mainly impacts unit tests)
Set this flag to true if local position is valid but accuracy low, such that
the operator can be warned before system switches to position-failure failsafe.
Additionally, switch to RTL if currently in Mission or Loiter to try to reach home
or fly out of GNSS-denied area.
Set low accuracy threshold to 50m by default for FW and VTOL.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Otherwise the flight time restriction flag gets cleared too early, before
disarming (which puts the vehicle into the previous mode and it might
take off again).
Fixes the following case:
- user intention set to X
- failsafe triggers, mode = Y
- can_run for X becomes false
- user tries to switch to X
-> need to re-evaluate can_run
Previously, it was not possible to enable forwarding of messages to/from
teh USB instance because it does not have a param for it, like the
serial instances have.
With this commit, we change the default to always set forwarding on for
the USB instance as that is likely desired.
Signed-off-by: Julian Oes <julian@oes.ch>
- from the C standard, "All identifiers that begin with an underscore and either an uppercase letter or another underscore are always reserved for any use"
- from the C standard, "All identifiers that begin with an underscore and either an uppercase letter or another underscore are always reserved for any use"
The static object is destroyed on at_exit while threads might still be
inside a CS. This can lead to a hanging process.
Cleaner would be to gracefully stop the threads.
According to https://linux.die.net/man/3/pthread_cond_destroy:
Attempting to destroy a condition variable upon which other threads are
currently blocked results in undefined behavior.
Instead check if system has previously switched into LOITER to acheive the current
WP of type POSITION, and in that case stay in LOITER until altitude is reached.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
- nuttx in PX4/Firmware (d3b7112dd94ac75a9a94ceef788d934f55c56c53): f80785664f
- nuttx current upstream: 35997053c5
- Changes: f80785664f...35997053c5
35997053c5 2023-01-23 David Sidrane - [BACKPORT] s32k1xx:serial Do not use TC use TDRE & TIE
Do the same as DO_REPOSITION wit only the altitude field populated
and MAV_DO_REPOSITION_FLAGS set, which means:
- switch to Loiter mode if not already in it
- set the current altitude to what is specified in the altitdue field,
keep current altitude setpoint otherwise
- keep current position setpoint
- fall back to current estimated position in case a position setpoint
is not finite
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
- sitl_gazebo-classic in PX4/Firmware (f3cdf70732c2b1ec24c8017aec1941e2295cdc7b): 1a725dd858
- sitl_gazebo-classic current upstream: 9343aaf4e2
- Changes: 1a725dd858...9343aaf4e2
9343aaf 2023-01-15 JaeyoungLim - Fix model prefix for user camera plugin (#948)
e5836d3 2023-01-13 Simone - Added robot namespace to the model (#941)
Previously the condition was based on hard coded height rate estimate and
setpoint values and an altitude error threshold. That showed to be leading
to false positives when the vehicle doesn't tightly follow the altitdue
ramp given by TECS to achieve a new altitude setpoint, and has become
completely infeasibly with the latest TECS rework that leads to non-ramped
altitude setpoint changes in the tecs_status message.
The new check no longer checks the altitude error but only the height rate
instead. It begins to integrate the height rate error once it detects an
uncommanded descend condition (height rate negative while setpoint is
positive). Integral threshold can be tuned by user (VT_QC_HR_ERROR_I).
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
In rare occasions asking for the protocol values after setting it returned
[0, 0]. I did not see any documentation for having to wait, but adding a
short wait period fixes it.
- refactor all EKF backend output predictor pieces into new OutputPredictor class
- output states are now calculated immediately with new high rate IMU rather than after EKF update
- IMU delayed sample is passed as around as control data to avoid storing an extra copy and make the requirement clear
- update motor mapping to use new UART_ESC_FUNC* auto generated params
- add support for actuator_test msg to support Actuator Testing in QGC
- modalai_fc-vX targets start driver if configured
- keep track of ESC spin direction in own param
- set ramp up param in MixerOutput to false
- use `gazebo-classic` everywhere consistently referring to the original Gazebo (eg version 9,10,11)
- additional `gazebo_*` helper targets added for compatibility, but warn about deprecation and tell you the new target naming
- use `gz` everywhere when referring to Gazebo (aka Ignition Gazebo or new Gazebo)
For the esc_calibration code, the actuator test is published for each motor
in a row. If the orb queue size is too small, messages are lost and not
received in mixer_module.
Set the default orb queue size of ActuatorTest high enough to keep the commands
for all motors in the queue.
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
Strictly follow the following convention for tailsitter:
FW Attitude and FW rate controller always operate in the FW frame, meaning that roll is
roll in FW, which for tailsitter means around the yaw axis in the body frame. The interfaces
between modules is though always in body frame.
That enables us to do the axis transformations for tailsitter, that are currently distributed
all over the controller (attitude, rate, vtol module), only at the input and output data of modules.
Side effect is that the FW rate control tuning gains meanings change: while before the roll gains
where meant for the body axis, they are now always applied for the FW roll axis (also in hover). So
the naming now is correct for FW, while before it was for Hover.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
This failsafe flag is currently used for not allowing to re-transition to FW, as well
as disabling pusher assist in hover. Till now it was only possible to reset it with
a commanded transition to MC, which many ground station interfaces didn't allow
as the system, after a quad-chute, already was in MC mode.
Hereby it is changed to reset when a new transition to FW is triggered (either via
RC switch or MAVLink command). It is the users responsibility to assess the situation
after a quad-chute happened to try to transition to FW again, manually proceed/land
the vehicle in MC, or let it finish the defined behavior after a quad-chute.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
- during casual testing on default configs the stack was penetration was reaching ~90% which is a bit too close for comfort
- increasing by 50% to be conservative
When we receive a command_long or command_int message to
answer, it arrives with a source sysid/compid, so this means we can send
the command_ack back on the appropriate MAVLink instances instead of all
of them.
This commit filters outgoing command_ack messages, so they are only sent
on the MAVLink instances where the sysid/compid has been seen in the
past.
This means that a command_ack is likely still sent on multiple links for
a setup with redundant links.
This should also prevent command_acks from being blasted on Iridium links
when it's not required.
Signed-off-by: Julian Oes <julian@oes.ch>
* Fixedwing rate control into a separate module
* Start fw_rate_control for vtol
* Move over airspeed related parameters to fw_rate_control
Co-authored-by: Silvan Fuhrer <silvan@auterion.com>
- libcanard in PX4/Firmware (280ef8935565d4a13e0780c926e1707fa5a90118): 2e3b11f6b8
- libcanard current upstream: 5c69d451ab
- Changes: 2e3b11f6b8...5c69d451ab
5c69d45 2022-11-03 John Vishnefske - Add sanitizers to unit test cmake build (#205)
989124d 2022-10-30 Pavel Kirienko - Fix 203 (#204)
f85103b 2022-10-13 joshvazquez-amzn - Fix warnings in test helpers (#202)
- sitl_gazebo in PX4/Firmware (bb591c6c29b49622bf79f2528b8382c930f3720e): 049b667d5e
- sitl_gazebo current upstream: d44ce17f43
- Changes: 049b667d5e...d44ce17f43
d44ce17 2022-12-19 Frederic Taillandier - fixing macos test build (#935)
7f9239c 2022-12-19 Daniel Mesham - Add safe landing world (#932)
61b1fc1 2022-12-19 Daniel Mesham - Update depth camera to match the RealSense D455 (#933)
On CubeOrange where no console is configured by default, starting
MAVLink shell just stalls, and doesn't work.
Also, logfile download has been reported not to work, and again, seems
to work with this change.
Signed-off-by: Julian Oes <julian@oes.ch>
- Do not pull in PWM parameters when DISABLE_PARAMS_MODULE_SCOPING is TRUE since VOXL2 has no PWM nor any of the required timer_config files that go along with that
- Replace non-standard M_PI constants with PX4 defined M_PI_F constants
- Include missing header file for function hrt_absolute_time declaration
- Add new PX4_SOC_ARCH_ID for the VOXL2 board
- move EV yaw and EV position to new state machines
- EV yaw and EV pos now configured via EKF2_EV_CTRL (migrated from EKF2_AID_MASK)
- new EV position offset estimator to enable EV position while GPS position is active (no more EV pos delta fusion)
- yaw_align now strictly means north (no more rotate external vision aid mask)
- automatic switching between EV yaw, and yaw align north based on GPS quality
* AirspeedSelector: use vehicle_air_data.rho for calculating groundspeed-wind CAS
Previously the vehicle_air_data.temperature and pressure was used, instead of the
density field directly.
Only makes a difference if there is an airspeed sensor connected to provide
the air temperature.
* AirspeedSelector: only safe estimated scale in param if airspeed is valid
* AirspeedSelector: remove 0.01 cliff for saving learned scale to param
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
- libevents in PX4/Firmware (26f3fea7ebb328ef58d4d592dae0559c91f13c1c): 0c8bc543db
- libevents current upstream: 8d9c555127
- Changes: 0c8bc543db...8d9c555127
8d9c555 2022-12-13 Beat Küng - README: clarify use of component ID
This checks was introduced to catch the (unlikely) case of the driver publishing
a stale value that is not actually from a current measurement right after boot.
This was done by declaring it invalid if there is no update of the measurement
within the first 10s after boot.
Practice though has shown that around 0, many airspeed sensors have such a low
resolution that the reported airspeed value is often at the exact same value
for longer periods of time on totally healthy sensors, and thus we trigger
some false positive failure detections.
Given that this failure mode (driver publishing stale data) is quite rare (if
it doesn't receive new data it should stop publishing), I remove this check
here again.
When in aerodynamic flight mode and armed, there is still the data stuck
check that can trigger if there is no update for 2s.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
* GPS_RAW_INT: Only send no gps messages if gps has ever been present
* GPS2_RAW: Keep in sync with GPS_RAW_INT
Signed-off-by: Marcell Rausch <marcell@auterion.com>
- respect new EKF2_EV_CTRL parameter for VPOS usage
- EV hgt rotate EV position before usage (there's often a small offset in frames)
- EV hgt reset use proper EV velocity body frame
- try to keep EV hgt and EV vel state machines consistent
- small incremental piece of https://github.com/PX4/PX4-Autopilot/pull/19128
- first version of IMU driver for the VOXL 2 platform (Qurt)
- this is a customized version of the Invensense ICM42688P driver, it is currently in the VOXL 2 board directory
- don't store unnecessary IMU copy in class, pass it through where required
- remove custom pi constants
- remove unused fields from ahrs_ekf_gsf (vel_NE, fuse_gps, accel_dt)
- explicitly initialize everything in header
- apply PX4 code style
- set GPS velocity before yaw estimator update, not after
- new ROS2 platform in PX4 intended for creating configs that build and run entirely in ROS2
- PX4_CONFIG defaults to px4_ros2_default if no config specified and in a colcon workspace with ROS_VERSION=2
- currently doesn't do much other than allow you to build px4 msgs interface package
Multiple agents can connect to the same client
For sitl builds, if the intance number is different from zero,
XRCE_DDS_KEY is set to the instance number and
the microdds_client is automatically started
with namespace
px4_"instance_number"
and udp port 8888
If the instance number is equal to zero
XRCE_DDS_KEY is left untouched and
the microdds_client is automatically started
without namespace
and udp port 8888
Signed-off-by: Beniamino Pozzan <beniamino.pozzan@phd.unipd.it>
The partecipant name is modified into
"client_namespace"/px4_micro_xrce_dds
For sitl builds the microdds_client is automatically started
with namespace
px4_"instance_number"
and udp port
8888+"intance_number"
Signed-off-by: Beniamino Pozzan <beniamino.pozzan@phd.unipd.it>
- sitl_gazebo in PX4/Firmware (b3cebc6686b0d8f86e4ecdc78c4b379509b510cb): b38e701ec4
- sitl_gazebo current upstream: 1b56329b73
- Changes: b38e701ec4...1b56329b73
1b56329 2022-11-22 Frederic Taillandier - adding a gazebo position sniffer to get position at a high rate from another program (#928)
- apps in PX4/Firmware (351bd1768a4a862f81b2a262225e945c0e4fe4fa): e04333c986
- apps current upstream: a489381b49
- Changes: e04333c986...a489381b49
a489381b4 2022-10-17 Peter van der Perk - [REJECTED] Backport 0d06c1c netinit:Network Monitor add a polled option
7b7cd332e 2022-09-12 Peter van der Perk - [REJECTED] Backport dbcb783671cee8a8c97b5909d9eb818c3864ca93 FAT DMA fix
With openjdk 17.0.5 2022-10-18, jmavsim fails to start due to missing
exports. This was previously a warning and turned into an error now.
Caught AppContextInfo(Bug 1004) InaccessibleObjectException: Unable to make public static sun.awt.AppContext sun.awt.AppContext.getAppContext() accessible: module java.desktop does not "exports sun.awt" to unnamed module @2c767a52 on thread J3D-Renderer-1
[0]: java.base/java.lang.reflect.AccessibleObject.checkCanSetAccessible(AccessibleObject.java:354)
[1]: java.base/java.lang.reflect.AccessibleObject.checkCanSetAccessible(AccessibleObject.java:297)
[2]: java.base/java.lang.reflect.Method.checkCanSetAccessible(Method.java:199)
[3]: java.base/java.lang.reflect.Method.setAccessible(Method.java:193)
[4]: com.jogamp.nativewindow.awt.AppContextInfo$1$1.run(AppContextInfo.java:40)
[5]: com.jogamp.common.util.UnsafeUtil.doWithoutIllegalAccessLogger(UnsafeUtil.java:219)
[6]: com.jogamp.nativewindow.awt.AppContextInfo$1.run(AppContextInfo.java:34)
[7]: java.base/java.security.AccessController.doPrivileged(AccessController.java:318)
[8]: com.jogamp.nativewindow.awt.AppContextInfo.<clinit>(AppContextInfo.java:31)
[9]: com.jogamp.nativewindow.awt.JAWTWindow.<init>(JAWTWindow.java:128)
[10]: jogamp.nativewindow.jawt.x11.X11JAWTWindow.<init>(X11JAWTWindow.java:60)
[11]: java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
[12]: java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:77)
[13]: java.base/jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
[14]: java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:499)
[15]: java.base/java.lang.reflect.Constructor.newInstance(Constructor.java:480)
[16]: jogamp.nativewindow.NativeWindowFactoryImpl.getAWTNativeWindow(NativeWindowFactoryImpl.java:105)
[17]: jogamp.nativewindow.NativeWindowFactoryImpl.getNativeWindowImpl(NativeWindowFactoryImpl.java:66)
[18]: com.jogamp.nativewindow.NativeWindowFactory.getNativeWindow(NativeWindowFactory.java:654)
[19]: javax.media.j3d.JoglPipeline$QueryCanvas.addNotify(JoglPipeline.java:8604)
[20]: java.desktop/java.awt.Container.addNotify(Container.java:2804)
[21]: java.desktop/java.awt.Window.addNotify(Window.java:791)
[22]: java.desktop/java.awt.Frame.addNotify(Frame.java:495)
[23]: java.desktop/java.awt.Window.show(Window.java:1053)
[24]: java.desktop/java.awt.Component.show(Component.java:1728)
[25]: java.desktop/java.awt.Component.setVisible(Component.java:1675)
[26]: java.desktop/java.awt.Window.setVisible(Window.java:1036)
[27]: javax.media.j3d.JoglPipeline.getBestConfiguration(JoglPipeline.java:8379)
[28]: javax.media.j3d.Renderer.doWork(Renderer.java:496)
[29]: javax.media.j3d.J3dThread.run(J3dThread.java:271)
Caught AppContextInfo(Bug 1004) IllegalAccessException: class com.jogamp.nativewindow.awt.AppContextInfo cannot access class sun.awt.AppContext (in module java.desktop) because module java.desktop does not export sun.awt to unnamed module @2c767a52 on thread J3D-Renderer-1
[0]: java.base/jdk.internal.reflect.Reflection.newIllegalAccessException(Reflection.java:392)
[1]: java.base/java.lang.reflect.AccessibleObject.checkAccess(AccessibleObject.java:674)
[2]: java.base/java.lang.reflect.Method.invoke(Method.java:560)
[3]: com.jogamp.nativewindow.awt.AppContextInfo.fetchAppContext(AppContextInfo.java:191)
[4]: com.jogamp.nativewindow.awt.AppContextInfo.update(AppContextInfo.java:135)
[5]: com.jogamp.nativewindow.awt.AppContextInfo.<init>(AppContextInfo.java:50)
[6]: com.jogamp.nativewindow.awt.JAWTWindow.<init>(JAWTWindow.java:128)
[7]: jogamp.nativewindow.jawt.x11.X11JAWTWindow.<init>(X11JAWTWindow.java:60)
[8]: java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance0(Native Method)
[9]: java.base/jdk.internal.reflect.NativeConstructorAccessorImpl.newInstance(NativeConstructorAccessorImpl.java:77)
[10]: java.base/jdk.internal.reflect.DelegatingConstructorAccessorImpl.newInstance(DelegatingConstructorAccessorImpl.java:45)
[11]: java.base/java.lang.reflect.Constructor.newInstanceWithCaller(Constructor.java:499)
[12]: java.base/java.lang.reflect.Constructor.newInstance(Constructor.java:480)
[13]: jogamp.nativewindow.NativeWindowFactoryImpl.getAWTNativeWindow(NativeWindowFactoryImpl.java:105)
[14]: jogamp.nativewindow.NativeWindowFactoryImpl.getNativeWindowImpl(NativeWindowFactoryImpl.java:66)
[15]: com.jogamp.nativewindow.NativeWindowFactory.getNativeWindow(NativeWindowFactory.java:654)
[16]: javax.media.j3d.JoglPipeline$QueryCanvas.addNotify(JoglPipeline.java:8604)
[17]: java.desktop/java.awt.Container.addNotify(Container.java:2804)
[18]: java.desktop/java.awt.Window.addNotify(Window.java:791)
[19]: java.desktop/java.awt.Frame.addNotify(Frame.java:495)
[20]: java.desktop/java.awt.Window.show(Window.java:1053)
[21]: java.desktop/java.awt.Component.show(Component.java:1728)
[22]: java.desktop/java.awt.Component.setVisible(Component.java:1675)
[23]: java.desktop/java.awt.Window.setVisible(Window.java:1036)
[24]: javax.media.j3d.JoglPipeline.getBestConfiguration(JoglPipeline.java:8379)
[25]: javax.media.j3d.Renderer.doWork(Renderer.java:496)
[26]: javax.media.j3d.J3dThread.run(J3dThread.java:271)
Exception in thread "main" java.lang.reflect.InvocationTargetException
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke0(Native Method)
at java.base/jdk.internal.reflect.NativeMethodAccessorImpl.invoke(NativeMethodAccessorImpl.java:77)
at java.base/jdk.internal.reflect.DelegatingMethodAccessorImpl.invoke(DelegatingMethodAccessorImpl.java:43)
at java.base/java.lang.reflect.Method.invoke(Method.java:568)
at org.eclipse.jdt.internal.jarinjarloader.JarRsrcLoader.main(JarRsrcLoader.java:61)
Caused by: java.lang.RuntimeException: java.lang.IllegalAccessException: class javax.media.j3d.JoglPipeline cannot access class sun.awt.X11GraphicsDevice (in module java.desktop) because module java.desktop does not export sun.awt to unnamed module @2c767a52
at javax.media.j3d.JoglPipeline.getScreen(JoglPipeline.java:8553)
at javax.media.j3d.Screen3D.<init>(Screen3D.java:354)
at javax.media.j3d.Canvas3D.<init>(Canvas3D.java:1124)
at javax.media.j3d.Canvas3D.<init>(Canvas3D.java:1026)
at javax.media.j3d.Canvas3D.<init>(Canvas3D.java:990)
at me.drton.jmavsim.Visualizer3D$CustomCanvas3D.<init>(Visualizer3D.java:909)
at me.drton.jmavsim.Visualizer3D.<init>(Visualizer3D.java:196)
at me.drton.jmavsim.Simulator.<init>(Simulator.java:193)
at me.drton.jmavsim.Simulator.main(Simulator.java:944)
... 5 more
Caused by: java.lang.IllegalAccessException: class javax.media.j3d.JoglPipeline cannot access class sun.awt.X11GraphicsDevice (in module java.desktop) because module java.desktop does not export sun.awt to unnamed module @2c767a52
at java.base/jdk.internal.reflect.Reflection.newIllegalAccessException(Reflection.java:392)
at java.base/java.lang.reflect.AccessibleObject.checkAccess(AccessibleObject.java:674)
at java.base/java.lang.reflect.Method.invoke(Method.java:560)
at javax.media.j3d.JoglPipeline.getScreen(JoglPipeline.java:8551)
... 13 more
* VTOL: take home_position into account for ground clearance
For approximating distance to ground in case there is no valid distance sensor
data, subtract the home position altitude from the local position altitude.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Co-authored-by: Matthias Grob <maetugr@gmail.com>
Instead of always starting with instance 0 (potentially an internal
mag), first take the current sensor as reference to compare the other
ones against it.
The issue is that otherwise we can end up in a
situation where a switch occurs because the currently used sensor isn't
much better than a sensor with a lower index: because the selected one
isn't much better, we cannot "fail-over" to it and get stuck on another
sensor, triggering a fail-over.
- WelfordMeanVector now computes covariance
- use Kahan summation for Welford mean (but continue using float32 for actual mean, etc)
- WelfordMean and WelfordMeanVector handle initial value and count roll over
- Welford mean count rollover at 16 bit max to prevent numerical issues and shift weight to newer samples
- sensors/vehicle_imu: update Welford mean usage (now simplified with resets removed)
- fix vehicle_imu_status accel var, now properly rotated with full covariance matrix
- gyro_calibration: update Welford mean usage
This is fully backwards compatible: If the throttle trim is set to
the minimum then it's the legacy calibration and gets
interpreted such that there is no trim and behavior remains as before.
If the trim is set to a different value than the minimum then it gets
used like with all other channels which was unsupported before.
In review it was requested to have a different name for
manual_control_setpoint.z because of the adjusted range.
I started to investigate what naming is most intuitive and found
that most people recognize the stick axes as roll, pitch, yaw, throttle.
It comes at no surprise because other autopilots
and APIs seem to share this convention.
While changing the code I realized that even within the code base
the axes are usually assigned to a variable with that name or
have comments next to the assignment clarifying the axes
using these names.
To be consistent with all other axes of stick input and avoid future
rescaling confusion.
Note: for the MAVLink message 69 MANUAL_CONTROL it's using the full range
according to the message specs now [-1000,1000].
Add ARK_FMU_V6X to RCS netman
Remove arkv6x rc single wire
Fix arkv6x mtd
arkv6x bootloader init all pins to prevent power cycling peripherals on boot
arkv6x don't power cycle sd card on boot
arkv6x add UART4 Telem 4
Also applies to Loiters that are started due to the previous mode being over (Takeoff,
VTOL_Takeoff, Mission).
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
landing airspeed was previously defined by a scale factor multiplied by minimum airspeed. this commit changes this parameter to an explicit speed, and when unspecified, defaults to the minimum airspeed
Use kClearanceAltitudeBuffer for it, which is also used to ensure that during takeoff an
altitude setpoint above the clearance altitdue is set.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
MIS_LTRMIN_ALt was used to limit the go-to altitude of a LOITER_TO_ALT (not the exit altitude,
but the altitude that the vehicle went to to fly to WP), and during landing abort to climb to
at least this altitude. The min entry altitude of LOITER_TO_ALT I remove with this commit, while
for the min alt during abort I added the new parameter MIS_LND_ABRT_ALT.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
previously, the next position setpoint in the triplet was left unchanged during an abort, which meant that the abort loiter current point combined with land next point triggered the early landing config logic. this commit is only a hack to make things work temporarily.. this needs to be handled better.
- explicitly defined takeoff airspeed setpoint
- dont use climbout mode
- allow max climb on takeoff
- dont handle post clearance altitude case (navigator will switch anyway)
Completely detach the steering wheel logic from the yaw controller (beside using the
same manual stick input in a manual flight mode).
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
- also put flaring internal states into a struct to organize a bit
- one concern with blending the throttle setpoint like this with the flare time param is that folding prop belly landing airframes may want to have a separate param for shorter throttle kill and still use the flare time ramps for everything else
abruptly changing to a heading setpoint on flare can cause the aircraft to roll and deviate from the runway, this commit
- maintains path following control during the flare not to disrupt the tracking just before touchdown
- (unfortunately for crosswind landing) removes the body axis alignment for runway bearing - this is a compromise
to achieve both runway bearing body axis alignment AND a specific touchdown point, either
1. the wind would need to be considered, and an appropriate diagonal approach (obstructions allowing) defined to the runway
2. slip control added, keeping path following outputs only commanding roll (controlling airspeed vector) and using yaw-rate command (only actuated by e.g. rudder) to align body axis with the runway
- consolidate takeoff rotation transition times for pitch constraints and throttle setpoint with a single param
- consolidate pitch takeoff constraint parameters (remove rwto_max_pitch, use nominal max)
- input correct units to rwto pitch constraint getters
- encapsulate absolute time interpolator method for transitions
- start runway ops from idle throttle
TECS climbout mode was used for takeoff climbout, which puts throttle to full and does not regulate a specific airspeed.
This commit sets the desired takeoff airspeed explicitly and allows max climb rate to track the ascent.
previously a scale factor param on min airspeed was used to define the climbout airspeed for runway takeoff
additionally, the rotation speed was defined by another hardcoded scale on top of the previously scaled min airspeed
this commit explicitly defines a takeoff speed and rotation speed for runway takeoff in params, with option to disable
Before, the logic to update disarm and rate values also triggered during
bootup on the px4io, because the output functions are only set in
updateSubscriptions().
Therefore change the check to prevent updating during the first cycle.
When using ignition SITL simulation with NO_LOCKSTEP, the SITL PX4 fails to update the IMU data from Ignition Gazebo.
The timestamp for the IMU data is taken from the ignition message:
- In LOCKSTEP mode the clock from the ignition simulation and the one from PX4 SITL are synchronized, hence everything works fine
- In NO_LOCKSTEP mode, those clocks are not synchronized anymore, so the timestamp for the IMU data should not be the one from Ignition but the current time in PX4 SITL when receiving the message.
The update function of InputMavlinkGimbalV2 returns UpdatedNotActive when receiving control commands from a system or component that doesn't match the primary control. This can happen if a component just sends commands without being in control or when transitioning to a different primary control.
If the input is marked as already_active, it will reset last_input_active which in turn resets the primary control in the next iteration. According to the MAVLink gimbal protocol v2, a component needs to send MAV_CMD_DO_GIMBAL_MANAGER_CONFIGURE to start controlling the gimbal and to remove control. However, with the current implementation there are several other commands that would reset the primary control.
With this PR, the primary control remains with the component that requested it last if updates from a not active component are received.
* Log position setpoint reference of npfg
This commit logs the local position setpoint reference when using NPFG
* Address review comments
This commit address review comments from @tstastny
Replace the existing check for the availability of a takeoff mission item with a combined
check for takeoff and landing item (or landing pattern). New param MIS_TKO_LAND_REQ
can be set to require only a takeoff, only a landing, both takeoff and landing, and
both or none. The latter is meant to be set if is e.g. deemed unsafe to start a flight
through a Takeoff WP without though defining a Landing - as then in case of a RTL the
vehicle doesn't follow a pre-defined path but instead only can do default RTL that especially
for FW and VTOL isn't always safe.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Co-authored-by: Peter van der Perk <peter.vanderperk@nxp.com>
Co-authored-by: David Sidrane <David.Sidrane@NscDg.com>
Co-authored-by: alexklimaj <alex@arkelectron.com>
Previously the minimum airspeed setpoint was adjusted to the load_factor compensated
stall speed, which, when the stall speed was set without margin, gave the controller
no room for error (the vehicle would stall if the controller has even a small airspeed
error).
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
- Include board_config.h for BOARD_GET_EXTERNAL_LOCKOUT_STATE etc. macros
- Include fcntl.h for "open"
Signed-off-by: Jukka Laitinen <jukkax@ssrc.tii.ae>
* LandDetectorMC: enforce that LNDMC_Z_VEL_MAX * 1.2 is below *MPC_LAND_CRWL/MPC_LAND_SPEED
Otherwise the _in_descend flag doesn't get set correctly during the last part
of the landing, where the descend speed is at MPC_LAND_CRWL or LAND_SPEED.
The _in_descend flag is set it the velocity setpoint is >1.1*LNDMC_Z_VEL_MAX.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Co-authored-by: Matthias Grob <maetugr@gmail.com>
* Made voxl2 apps processor and slpi dsp processor builds into separate board builds so that they can
more easily be configured independently.
* Removed board specific link library command from px4_config.cmake and moved it to a more generic
board specific solution that can be used by any board that needs custom link libraries.
* Removed redundant cmake command for Qurt
* Removed unused definition from Qurt cmake file
* Removed unnecessary QURT_LIB cmake function
* Reorganized the voxl2 build structure to avoid 4 level board directories.
* Reverted cmake files to remove 4 level board naming code
* Updated documentation
- in a lot of cases this won't really matter, although you can see it in logs (especially with things like vehicle_at_rest), but it's something we might as well do properly
Co-authored-by: Mathieu Bresciani <brescianimathieu@gmail.com>
Fixes a regression from 8bae4e5c0e29f70d5d1d2427ffdef97092be939c, where
the orbit flight task wasn't an extra task (flight_tasks_to_add) anymore
and therefore the command handling wasn't generated.
There was a race condition that could cause several outcomes. The most severe
was that flight_mode_manager gets the command, switches to orbit and then
in the next iteration switches back because commander did not change
nav_state yet. When commander then switches, flight_mode_manager would still
be in the old mode.
This is prevented by storing the command (allowing it to arrive before or
after mode switch), and then apply it after the switch happens.
- vehicle_angular_velocity and vehicle_angular_acceleration are produced together from the same input data, consumed together, and share the the same metadata (timestamps)
- individually these topics each have 16 bytes of metadata (2 timestamps) for 12 bytes of data (x,y,z float32)
During the whole flight, if the difference between the yaw estimate from
EKF2 and the emergency estimator is large and that the GNSS velocity
fusion is failing continuously, immediately reset to the emergency yaw
estimate.
- with Ekf::initialiseFilter() resetting _time_last_hagl_fuse the terrain estimate was technically valid regardless of any range or flow data availability and as a result optical flow fusion is able to start and stay active
- only consider terrain estimate valid based on control status flags and recent fusion
The CameraFeedback module used only the vehicle attitude for the camera orientation so far. With this change, the gimbal_device_attitude_status is used to compute the global camera orientation when a gimbal is used.
In this case it did not cause any problems.
Fixes a compiler warning:
/src/modules/commander/HealthAndArmingChecks/HealthAndArmingChecks.cpp:39:21: error: member ‘HealthAndArmingChecks::_failsafe_flags’ is used uninitialized [-Werror=uninitialized]
39 | _reporter(_failsafe_flags)
| ^~~~~~~~~~~~~~~
* simulator_mavlink: Add basic vio failure injection
Signed-off-by: Paul Frivold <paul@kefrobotics.com>
* simulator_mavlink: Rm failure not supported warning
Failures commands are also handled in other files,
so warning here could be confusing.
Signed-off-by: Paul Frivold <paul@kefrobotics.com>
Signed-off-by: Paul Frivold <paul@kefrobotics.com>
previous change in logic to hold after mission clear also broke rtl, as non-mission takeoff still published a mission result which allowed entering the mission finished condition and always changing state to loiter (ignoring rtl). new logic only switches navigation states if mission is finish and the nav state is explicitly in takeoff state, or in mission state
The srcparser.py is specific to each use case (e.g. Airframes, Parameters, px4events, etc as in Tool/* folders).
Therefore it is confusing to have the px_process_airframes.py script handle concept of airframes under the generic name 'params'.
This improves readability and sets the baseground for implementing more specific vehicle type supports, as mentioned in https://github.com/PX4/PX4-user_guide/pull/1858#discussion_r876554728
once the rampup starts. The rampup requires a valid vertical velocity setpoint.
The corner case is:
- We are before takeoff and amending the setpoint to be 0,0,100 acceleration
in order to idle
- The rampup starts BUT the setpoint is not yet overwritten by the trajectory
setpoint topic
- The idle setpoint gets amended to not contain a feed-forward vertical
acceleration because the rampup is velocity based
- The result is a brief invalid 0,0,NAN acceleration setpoint
- That invalid setpoint gets overridden by a failsafe that holds zero velocity
- Zero velocity leads to applying ~hover thrust briefly
- buffer at least 2 samples for the IMU output predictor buffers
- dropping below 2 becomes problematic for the minimum observation
interval calculation and the vertical output buffer trapezoidal
integration
- sitl_gazebo in PX4/Firmware (0e2b1ee9798fc2662dbfdaca68a5a8ae9997b28d): 56b5508b72
- sitl_gazebo current upstream: b38e701ec4
- Changes: 56b5508b72...b38e701ec4
b38e701 2022-11-03 frederictaillandier - adding jinja parameters
cc8f33d 2022-11-03 frederictaillandier - allowing to override the streaming gimbal ip and port
6c3f1f8 2022-11-03 alessandro - send orientation q with landing_target message (#919)
48e764b 2022-11-02 Frederic Taillandier - allowing to override the video stream ip in a simpler way than using the jinja file (#923)
arm
- Previously, due to the way MAVSDK's `health_all_ok` was implemented,
vehicle often didn't have a valid global position estimate (although
function returned true), and it wouldn't arm, and the SITL would fail
- Also sometimes as vehicle didn't have manual control, it entered weird
states where it wasn't able to arm as well
- This adds a check to make sure vehicle is able to arm, directly from
the Health struct
- preflight mag bias estimate is in the vehicle body frame and removed from the raw mag data after it's rotated, calibrated, etc
- in flight mag bias (from ekf2) is in the vehicle body frame, but stored as a sensor frame offset immediately
* Enable motor controls for fixed wing mode in tailsitters
This commit enable motor controls in fixed wing mode for tailsitters
This is needed for enabling quad tailsitters
* VTOL: differential thrust in FW: adapt params to be generic for all axes
Until now only suppoted on yaw axis. Not to be supported also on Roll and Pitch.
- VT_FW_DIFTHR_EN: make bitmask for all three axes independently. First bit is Yaw,
sucht that existing meaning of VT_FW_DIFTHR_EN doesn't change.
- VT_FW_DIFTHR_SC: rename to VT_FW_DIFTHR_S_Y and add same params for roll (_R) and
pitch (_P).
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
* Integrate differential control bits to three axis control
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
It's enough that the setpoints and the unallocated values are logged, from these
the allocated values can be calculated if required.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
It's very important that the in descend detection is always
at a strictly higher velocity than the vertical movement check.
This combination is basically used to check for vertical downwards
velocity tracking. Desired descend, no movement -> ground
If in descend threshold is lower than vertical movement it is
by definition even with perfect tracking the case that with
any velocity between the two thresholds there is no movement
even though a descend is commanded.
See first fix of this problem #7831
e39b38ba96971245aaf6d2b1c249868c8717665e
I don't see where this is necessary. During takeoff, the maybe landed flag should
only get cleared once system is about to takeoff, and thus well after the spool up
is complete (for which the higher thresholds are meant in this case).
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Don't consider these params for vertical speed threshold,
and instead increase the default for LNDMC_Z_VEL_MAX and
use solely that one. Makes the land detector outcome more
predictable and its interal logic simpler, while the new
default tuning is resulting in about the same vz threshold
as before.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
* Changed M_PI to M_PI_F in the matrix library since M_PI is non-standard.
* Added a new M_PI_PRECISE constant definition to px4_platform_common/defines.h to be
used in places when M_PI is desired but shouldn't be used because it is not C standard.
* Added the px4_platform_common/defines.h include to the matrix library math.hpp header to pull
in some non-standard M_PI constants and updated the test files to use those constants.
* Fixed PI constants in matrix helper test to prevent test failure
- GPS drivers in PX4/Firmware (d67b19ac1d41b2dcfc61ed6d353ae513ac3f4a82): 1ff87868f6
- GPS drivers current upstream: fa2177d690
- Changes: 1ff87868f6...fa2177d690
fa2177d 2022-10-10 Michael Schaeuble - Return from GPSDriverUBX::receive when ready, don't wait until no data is received
Co-authored-by: PX4 BuildBot <bot@px4.io>
before there was a corner case where if in an auto mode that sets a reset command on the attitude setpoint message (e.g. auto takeoff), if the mode was then switched stabilized, this reset bool would never be changed back to false and the integrators would reset every cycle
* Removed dprintf from perf library since it is only ever used with fd=1 (STDOUT) so moved to PX4_INFO_RAW instead. This helps with some platforms (e.g. Qurt) which have some Posix support but not full Posix support.
This was previously required to reset the flight speed after a VTOL transition,
but is now no longer required as the DO_CHANGE_SPEED commands are handles directly
in the controllers.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
This fixes the issue where the init happended in the initializer list, at which point
the params were not yet initialized and thus resulted in random values for the init
values of _mission_item.loiter_radius and .acceptance_radius.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Refactor
- Require reboot for PD_GRIPPER_EN parameter change
- Define gripper ACTION_NONE for readability. This makes implicit assumption that -1 equals no-action commanded more explicit
- Tidy the scattered vcmd_ack struct handling cases into a single function
- Refactor to remove return in the middle of function: avoids future complications where a programmer may expect the logic at the end of the function to be executed, but isn't
Vehicle Command Handling
- Cancel the previous running vehicle command if we receive a different vehicle command
- Reject vehicle command if we get a same one that is getting executed
- Save the source system & component of currently running vehicle command
- Added note about new discovered edge case of having same entity sending different gripper commands consequently, where an unexpected ack result may be received
Gripper:
- Don't command gripper (via uORB `gripper` topic, which maps into an
actuator via Control Allocation) if we are already at the state we want
(e.g. Grabbed / Released) or in the intermediate state to the state we
want -> This prevents spamming on `gripper` topic
Payload Deliverer:
- Add read-once function for Gripper's released / grabbed state
- Send vehicle_command_ack for both release/grab actions.
TODO: target_system & target_component for the released/grabbed vcmd_ack
is incomplete, since we are not keeping track of the vehicle_command
that corresponds to this. This needs to be dealt with in the future, not
sure what the best solution it is for now.
Possible solutions:
- Queue-ing the vehicle command?
- Tying the gripper's action to specific vehicle command one-on-one, to make sure if we send multiple vehicle commands, we know
which command resulted in the action exactly?)
Only command Gripper grab when we are actually initializing gripper
- Previously, on every parameter update, gripper grab was being
commanded
- This commit narrows that scope to only when we are actually
initializing the gripper
Handle gripper de-initialization upon parameter change
- Also added some local state initialization code to init() function of
Gripper
- This will now make init / de-init more graceful & controlled compared
to before
- This hopefully then alerts the GCS that the command is getting
processed
- Referenced commander's `handle_command` function to implement this. As
it seems like GCS needs the acknowledgement of the command being
processed to execute such commands properly
- Also send FAILED command ack if we can't actuate the gripper
Fix wrong GRIPPER_ACTION conversion from floating point to int32_t
- Due to the MAVLink spec, we actually just convert enums into floating
point, so in PX4 we need to convert the float directly into integer as
well (although there can be precision issues on large numbers)
- This is a limitation in MAVLink spec, and should hopefully be
changed in MAVLink v2
Affects the states AUTO_ALTITUDE and AUTO_CLIMBRATE. Those modes should only be entered
if armed (as they are pure failsafe modes). Also allow though to enter them even if
the position setpoint(s) are invalid, as they are not needed.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
Instead of checking the .valid flag of position_setpoint, check for ISFINITE() of lat, lon, alt
when pulling the position_setpoint triplet. This fixes problems where the .valid flag didn't
reflect the proper state of the setpoint (e.g. .valid was true, .lat though NAN)
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
the nuttx and posix specific options files since this option cannot be used with
the qurt platform. There are header files in the hexagon sdk that fail this check.
as Tools/update_px4_ros2_bridge.sh as been deleted
update_ros2_bridge, update_px4_ros_com and update_px4_msgs
are no more needed
Signed-off-by: Beniamino Pozzan <beniamino.pozzan@phd.unipd.it>
- vehicle_angular_velocity: ESC RPM notch filters minimize filter resets
- only allow one filter init per axis per cycle
- "park" ESC notch filters at min frequency instead of full disable
- relax timeout before a notch filter is disabled
- add new parameter IMU_GYRO_DNF_MIN for configuring the minimum notch filter frequency
- avoid the possibility of unsigned underflow from subtracting two HRT timestamps (uint64_t)
- most of these aren't problematic, but people tend to replicate the pattern, so it's better to be safe
- likely wasn't a problem when people were using hrt_absolute_time() in place, but if using an existing timestamp there's the possibility it's older than a more recent topic update
- update all msgs to be directly compatible with ROS2
- microdds_client improvements
- timesync
- reduced code size
- add to most default builds if we can afford it
- lots of other little changes
- purge fastrtps (I tried to save this multiple times, but kept hitting roadblocks)
- working towards keeping all height source (baro/ev/gnss/rng) handling as consistent as possible, possibly refactoring these out into separate classes later
This fixes a bug where by accident the vtol_status was considered instead of the
vehicle_status, preventing it from running on planes.
Signed-off-by: Silvan Fuhrer <silvan@auterion.com>
2022-10-19 10:35:28 +02:00
2102 changed files with 118893 additions and 53505 deletions
Please use [PX4 Discuss](http://discuss.px4.io/) or [Slack](http://slack.px4.io/) to align on pull requests if necessary. You can then open draft pull requests to get early feedback.
<!--
## Describe problem solved by this pull request
A clear and concise description of the problem this proposed change will solve. Or, what it will improve.
E.g. For this use case I ran into...
Thank you for your contribution!
## Describe your solution
A clear and concise description of what you have implemented.
Get early feedback through
- Dronecode Discord: https://discord.gg/dronecode
- PX4 Discuss: http://discuss.px4.io/
- opening a draft pr and sharing the link
## Describe possible alternatives
A clear and concise description of alternative solutions or features you've considered.
-->
## Test data / coverage
How was it tested? What cases were covered? Logs uploaded to https://review.px4.io/ and screenshots of the important plot parts.
See also [maintainers list](https://px4.io/community/maintainers/) (px4.io) and the [contributors list](https://github.com/PX4/PX4-Autopilot/graphs/contributors) (Github).
Blocking a user prevents them from interacting with repositories, such as opening or commenting on pull requests or issues. Learn more about blocking a user.
