Add camera zoom plugin

[srmainwaring]

Add a 3 axis gimbal and system plugin that attaches to a camera sensor and allows the camera to be zoomed.

Details

• Add 3 axis version of the small gimbal.
• Add example world with mount and objects of interest.
• Add system plugin that subscribes to a zoom command topic (msgs::Double) and allows an existing camera sensor to be zoomed.

The plugin SDF has a single parameter setting the maximum zoom factor.

 <sensor name="camera" type="camera">
   ...
   <plugin filename="CameraZoomPlugin"
       name="CameraZoomPlugin">
     <max_zoom>15.0</max_zoom>
   </plugin>
</sensor>

Testing

camera_zoom.mov

The gimbal is position controlled and responds to roll, pitch and yaw commands.

gz topic -t "/gimbal/cmd_yaw" -m gz.msgs.Double -p "data: -1"

The camera zoom is similarly commanded.

gz topic -t "/model/gimbal/sensor/camera/zoom/cmd_zoom" -m gz.msgs.Double -p "data: 2"

Tasks

• #include <gz/sim/Sensor.hh> is not available in the released version of gz-sim7 on Ubuntu.
• Resolve segmentation fault on exit.

[Ryanf55]

Have you considered specifying zoom as a focal length? It's more deterministic.

[srmainwaring]

Yes, could do. This is just a basic first pass to get the internal mechanics of the plugin working and a factor applied to HFOV was the method used in the PX4-SITL_Gazebo plugin for Gazebo Classic so I took my steer from that.

[rmackay9]

It's really great to see this, thanks so much!

[Ryanf55]

If you are still interested to get this merged, let me know! I think it's a great feature addition.

[srmainwaring]

If you are still interested to get this merged, let me know! I think it's a great feature addition.

Thanks @Ryanf55, let me re-test on Harmonic. IIRC there was an issue in Gazebo Garden that could cause crashes and I need to remind my self what it was and whether it has been fixed.

[srmainwaring]

If you are still interested to get this merged, let me know! I think it's a great feature addition.

Hi @Ryanf55 - tested on Harmonic and appears to now be working fine. I can not longer see a segmentation fault on exit, so I assume the issue (which was connected to the physics libraries unloading IIRC) has been resolved upstream.