Buoyancy plugin

[acxz]

Solves #484 and #167

The buoyancy computation code is removed from usv_dynamics and the generic BuoyancyPlugin is used.

The buoyant force has been verified to be within 50 N of the previous code. Let me know if you would like videos for verification.

[bsb808]

@acxz This is a really interesting idea. When we wrote this plugin the standard Gazebo buoyancy plugin wasn't able to deal with items at the surface - only fully submerged items.

I tested the modification with an exaggerated sea-state. There seems to be issues with synchronization with the visual wave field and lack of induced vessel pitch and roll. Those are critical capabilities for our use-case.

I'm not very familiar with the latest buoyancy. Perhaps the standard plugin can be used for this application. Is there documentation on how the plugin uses the <wave_model> and "buoyancy" tags?

[bsb808]

[acxz]

@bsb808 thanks for looking at this PR!

I tested the modification with an exaggerated sea-state.

Can you provide the files you changed, so that I can test the scenario on my local machine.

There seems to be issues with synchronization with the visual wave field and lack of induced vessel pitch and roll.

This is a problem of the wave field visuals vs the actual simulated wave field height. You will also see similar behavior with the existing buoyancy code, without the plugin. Have you compared the two in your exaggerated scenario? I ran my own exaggerated scenario setting scale to 1000 here

vrx/wave_gazebo/world_models/ocean_waves/model.xacro

Line 42 in e5544ed

<scale>${scale}</scale>

and here

vrx/wave_gazebo/world_models/ocean_waves/model.xacro

Line 63 in e5544ed

<scale>${scale}</scale>

and observed the same behavior. A different issue should be opened for the discrepancy between the wavemodel and wavevisuals.

As for the lack of induced pitch and roll, the previous code also does not handle it. This PR is only meant to be a generalization to vessel's buoyancy computation, instead of hardcoding it as two half cyclinders. A different issue should be opened for the lack of induced pitch and roll.

Perhaps the standard plugin can be used for this application. Is there documentation on how the plugin uses the <wave_model> and "buoyancy" tags?

Here is the current upstream BuoyancyPlugin code: https://github.com/osrf/gazebo/blob/2e42eb91e7638da54b36dbb8fdd4ae5e3a310d6d/plugins/BuoyancyPlugin.cc

As you can tell, the upstream BuoyancyPlugin is not at all as fancy as the one we have here in VRX. I don't think this is an option when trying to improve the buoyancy code we have in VRX.

It also does not use the <wave_model> tag.

[bsb808]

See https://github.com/osrf/vrx/wiki/changing_plugin_params_tutorial to change wave parameters. I probably used period=10 and gain=1

[acxz]

Thanks for sending the params you tested on. Just tested this PR and the existing codebase and the same behavior is observed.

For this PR in particular is there any closure that I can expect, now that the master branch has been put in maintenance mode.

cc'ing @M1chaelM since he has made issue #487 which this PR helps tackle.