gz: fix gimbal yaw, add dds publisher

[dakejahl]

In conjunction with PX4/PX4-gazebo-models#122

context
Support for using x500_gimbal in our px4-ros2 precision landing example.

[github-actions]

🔎 FLASH Analysis

px4_fmu-v5x [Total VM Diff: 480 byte (0.02 %)]

    FILE SIZE        VM SIZE    
--------------  -------------- 
+0.0%    +480  +0.0%    +480    .text
  [NEW]    +288  [NEW]    +288    ucdr_serialize_gimbal_device_attitude_status()
  +0.1%    +120  +0.1%    +120    [section .text]
  +2.6%     +84  +2.6%     +84    UxrceddsClient::run()
  +5.0%      +4  +5.0%      +4    write_user_stack.constprop.0
 -11.1%     -16 -11.1%     -16    px4::wq_configurations::lp_default
+0.0%     +55  [ = ]       0    .debug_abbrev
+0.0%      +8  [ = ]       0    .debug_aranges
+0.0%     +28  [ = ]       0    .debug_frame
+0.0%    +704  [ = ]       0    .debug_info
+0.0%    +617  [ = ]       0    .debug_line
+0.0%     +73  [ = ]       0    .debug_loclists
-0.0%     -20  [ = ]       0    .debug_rnglists
  [DEL]      -3  [ = ]       0    [Unmapped]
  -0.0%     -17  [ = ]       0    [section .debug_rnglists]
+0.0%    +255  [ = ]       0    .debug_str
-1.2%      -3  [ = ]       0    .shstrtab
+0.0%     +67  [ = ]       0    .strtab
  [NEW]     +67  [ = ]       0    ucdr_serialize_gimbal_device_attitude_status()
+0.0%     +32  [ = ]       0    .symtab
  [NEW]     +32  [ = ]       0    ucdr_serialize_gimbal_device_attitude_status()
-5.4%    -480  [ = ]       0    [Unmapped]
+0.0% +1.77Ki  +0.0%    +480    TOTAL

px4_fmu-v6x [Total VM Diff: 480 byte (0.02 %)]

    FILE SIZE        VM SIZE    
--------------  -------------- 
+0.0%    +480  +0.0%    +480    .text
  [NEW]    +288  [NEW]    +288    ucdr_serialize_gimbal_device_attitude_status()
  +0.1%    +104  +0.1%    +104    [section .text]
  +2.6%     +84  +2.6%     +84    UxrceddsClient::run()
   +44%      +4   +44%      +4    g_nullstring
  +5.0%      +4  +5.0%      +4    write_user_stack.constprop.0
  -4.5%      -4  -4.5%      -4    FlightTask
+0.0%     +55  [ = ]       0    .debug_abbrev
+0.0%      +8  [ = ]       0    .debug_aranges
+0.0%     +28  [ = ]       0    .debug_frame
+0.0%    +704  [ = ]       0    .debug_info
+0.0%    +617  [ = ]       0    .debug_line
+0.0%     +73  [ = ]       0    .debug_loclists
-0.0%     -16  [ = ]       0    .debug_rnglists
   +50%      +1  [ = ]       0    [Unmapped]
  -0.0%     -17  [ = ]       0    [section .debug_rnglists]
+0.0%    +255  [ = ]       0    .debug_str
+0.4%      +1  [ = ]       0    .shstrtab
+0.0%     +67  [ = ]       0    .strtab
  [NEW]     +67  [ = ]       0    ucdr_serialize_gimbal_device_attitude_status()
+0.0%     +32  [ = ]       0    .symtab
  [NEW]     +32  [ = ]       0    ucdr_serialize_gimbal_device_attitude_status()
-10.5%    -480  [ = ]       0    [Unmapped]
+0.0% +1.78Ki  +0.0%    +480    TOTAL

Updated: 2025-10-14T18:50:20

[beniaminopozzan]

@dakejahl , in

matrix::Quatf q_yaw_correction = matrix::Quatf(cosf(yaw_offset / 2), 0, 0, sinf(yaw_offset / 2));

matrix::Quatf q_gimbal_FLU = q_yaw_correction * q_gimbal_FLU_raw;
_q_gimbal = q_FLU_to_FRD * q_gimbal_FLU * q_FLU_to_FRD.inversed();

you're computing

_q_gimbal = q_FLU_to_FRD *  q_yaw_correction * q_gimbal_FLU_raw * q_FLU_to_FRD.inversed();

There q_FLU_to_FRD * q_yaw_correction is a ENU to NED rotation so I pushed a simplified version of the formula.

Would you be able to test it please?

Nevertheless, the PR looks goods, if you prefer to keep your version feel free to force push it, I'll approve straight away

[dakejahl]

@beniaminopozzan thank you for the simplification! I just tested and it works the same as before