* [ocaml] apply a second topological sort to modules
After the first one, all functionalities are checked, but corresponding
modules are not ordered with the other one.
This second sorting algorithm (simpler) is just checking that final
selection of modules is correctly ordered and check for possible cyclic
dependencies not yet found.
ahrs_common dep node is fixed to avoid such a situation.
* [modules] trying to fix cyclic dependency in rotorcraft gnc modules
the main issue is that the rotorcraft navigation is not explicitely
included by the user but automagically included, which is not a good
idea, same goes for guidance.
* [modules] fix dep
system_core is already provided by firmware module and @no_settings
was making a cyclic dep for fbw
* [server] stream telemetry messages as JSON string
can be used by external plotter like PlotJuggler (default port)
always enabled by default
* update pprzlink
Co-authored-by: Fabien-B <fabien.bonneval@gmail.com>
* [ocaml] drop of compat for Ocaml < 4.05
this is linked to Ubuntu 16.04 that have reached end of life
* [pprzlink] update pprzlink for removed old compat
This commit implements tilt twist control for vehicles that rotate slow
in the z axis compared to the other axes (like tailsitters).
The attitude error is decomposed in a tilt and a twist part, and the
tilt error is compensated assuming that the rotation around z is zero
(very slow in comparison).
* [server] Make "time_since_last_msg" of TELEMETRY_STATUS link specific.
* [GCS] Handle link specific "time_since_last_msg".
* Fix for old OCaml compatibility.
* Fix minimum last_msg time.
* Update link_id when no_id is replaced.
daba128 changed the Dockerfile to install gosu from the official Ubuntu repositories but did not change the path in entrypoint.sh
gosu ist now located at /usr/sbin/gosu and no longer at /usr/local/bin/gosu
- try to set the sensor frequency automatically at the correct value
- make more parameters configurable
- set higher bounds for all sensors to avoid saturation
- remove old unused files
- update airframe files
It prevents the filter to jump to the new position, which can create a
wrong estimation of the accelerometer scale bias.
Aligner is restarted if needed.
Also fix float constants and other small details.
- with ChibiOS 20.3 the half buffer starting position is computed by
hand and was wrong in this case: it should be the product of the half
buffer size by the number of channels
- some other minor fixes:
- using && instead of & in boolean comparison
- suppress a warning (comparison always true/false due to limited data range)
when the number of channels is zero (no ADC in use)
Gautier Hattenberger
Hector Garcia de Marina
Prawien Kanhai
Matteo Barbera
Ewoud Smeur
Fabien-B
Pietro Campolucci
amtcvx
winkelst
Freek van Tienen
OpenUAS