underwater robot/uuv extension #402

DanielDuecker · 2020-01-31T16:51:09Z

This PR corresponds to PX4/PX4-Autopilot#14079

It extends the underwater robot capabilities and provides an uuv_motor_model allows to model the esc-deadzone which is important for vehicle actuators which can spin in both directions.

The old hippocampus model is kept for the moment as part of the px4-firmware tutorial depends on it. However it likely makes sense to have the tutorial depending on the new model at some point.

Merge branch 'hippocampus' of https://github.com/zarbokk/sitl_gazebo into uuv_extension

* general code clean up + comments

…_gazebo into uuv_extension

…extension

Jaeyoung-Lim

Thanks for the contribution! I have a few general comments as the following.

The name of the plugin is gazebo_uuv_motor_model. However, what the plugin is actually doing is simulating the uuv dynamics. It includes the combined dynamics of 4 motors and the damping dynamics of the body. I think this is confusing, since the gazebo_motor_model for all the rest of the models only include the dynamics for the motor itself. For plane / multirotos there are separate plugins for the motor model and the fuselarge interaction with aerodynamics, and multiple motor elements are added depending on how many actuators there are in the system.
This is more out of curiousity, as the thrust is modeled linear to the rotational velocity. Could you point us to any reason / literature that this was modeled this way? I am not an expert of UUVs, but I would expect that in a newtonian fluid (such as water) the thrust will increase quadratically compared to the rotational velocity of the srews. Does the linear model better fit in low thrust scenarios for the hippocampus?

src/gazebo_uuv_plugin.cpp

models/uuv_hippocampus/uuv_hippocampus.sdf

Jaeyoung-Lim · 2020-02-03T10:16:49Z

models/uuv_hippocampus/uuv_hippocampus.sdf

+    <plugin name="gps_plugin" filename="libgazebo_gps_plugin.so">
+        <robotNamespace/>
+        <gpsNoise>true</gpsNoise>
+    </plugin>


Does the hippocampus have GPS?

Our real world version does not have GPS since GPS does not work/is attenuated almost immediately underwater. Underwater localization is some kind of a night mare which I would prefer for the moment to keep out of the simulation
Hence we implemented "GPS" as a well integrated work around with fairly similar characteristics. However, this can be of course extended at some point but it will be quite a chunk of work to model the real world localization characteristics.

Fair enough. I don't think we need to "fix" this, but still be aware the state estimation is using GPS data

* updates on the new uuv_hippocampus model - add nicely spinning propeller - add red/green coloring to propellers -> port/starboard side * Merge the new uuv_motor_model into the UUV_plugin, which now - computes the hydrodynamic effects (according to FOSSEN 2011 - Handbook of Marine Craft Hydrodynamics and Motion Control) - applies the Hydro effects as forces/moments and are than applied to the vehicle - computes the forces/moments based on the motor commands

DanielDuecker · 2020-02-04T14:17:28Z

Thanks for the contribution! I have a few general comments as the following.

The name of the plugin is gazebo_uuv_motor_model. However, what the plugin is actually doing is simulating the uuv dynamics. It includes the combined dynamics of 4 motors and the damping dynamics of the body. I think this is confusing, since the gazebo_motor_model for all the rest of the models only include the dynamics for the motor itself. For plane / multirotos there are separate plugins for the motor model and the fuselarge interaction with aerodynamics, and multiple motor elements are added depending on how many actuators there are in the system.

This is more out of curiousity, as the thrust is modeled linear to the rotational velocity. Could you point us to any reason / literature that this was modeled this way? I am not an expert of UUVs, but I would expect that in a newtonian fluid (such as water) the thrust will increase quadratically compared to the rotational velocity of the srews. Does the linear model better fit in low thrust scenarios for the hippocampus?

@Jaeyoung-Lim

with the last commit I merged the uuv_motor_model back into the uuv_plugin (I agree this makes much more sense than keeping them separate)
You are right with the thrust modelling. However, it is actually modeled quadratically, see uuv_plugin.cpp , ca line 155:
forces[i] = motor_force_constant_ * motor_commands_[i] * std::abs(motor_commands_[i]);

Jaeyoung-Lim · 2020-02-04T16:06:22Z

@DanielDuecker For 1. and 2, Ah, but if the thrust is quadratic, why do we need an additional plugin to model the dynamics of the thrusters? I understand that the lift / drag would be very different since their is the effect of added mass since the fluid is incompressible.

What are the fundamental differences between the underwater thrusters and rotors in the air? If the differences are not that big, Wouldn't it make sense that we use the existing motor plugin? This would make a much cleaner and scalable solution where you can add arbitrarily many thrusters.

DanielDuecker · 2020-02-04T16:33:52Z

@Jaeyoung-Lim
so far we just use quadcopter motors underwater which works surprisingly well.
A more detailed analysys of the single motor dynamics would be definitely helpful. Unfortunately, I cannot provide detailed parameters on this for the moment.

Therefore, it is currently easier to model the overall vehicle. Moreover, currently the main point of the uuv_plugin is the extension to added mass and added damping of the overall vehicle. Hereby, we made the experience that even the mounting of the motors (outside the vehicle, see photo) does have quite some effect on the added mass/damping - I think it would be great, but quite tough, to have this in a fully modular concept.

Jaeyoung-Lim · 2020-02-04T16:35:31Z

@DanielDuecker That's fair. Just wanted to make sure that it is necessary to include the motor models in the plugins

DanielDuecker · 2020-02-05T07:57:28Z

@dagar @Jaeyoung-Lim
what else do you think should be done here before merging?
It would be great to have this merged soon, as the tutorial's command make px4_sitl gazebo_uuv_hippocampus otherwise throws an error due to the missing sitl_gazebo connection

Jaeyoung-Lim

I am not really happy with the fact that the thrusters are included in the plugin and that you are not able to add more thrusters.

However, since this is specific to the hippocampus and @DanielDuecker is developing it, I think this is good to go on myside.

- major renaming from uuv2 -> uuv_hippocampus - added uuv_hippocampus model carrying a camera - added uuv_hippocampus_cam.world - updates on the new uuv_hippocampus model - add nicely spinning propeller - add red/green coloring to propellers -> port/starboard side - computes the hydrodynamic effects (according to FOSSEN 2011 - Handbook of Marine Craft Hydrodynamics and Motion Control) - applies the Hydro effects as forces/moments and are than applied to the vehicle - computes the forces/moments based on the motor commands Co-authored-by: PhilippHastedt <34442705+PhilippHastedt@users.noreply.github.com> Co-authored-by: Tim <t.hansen@tu-harburg.de>

PhilippHastedt and others added 15 commits November 1, 2019 22:37

Add UUV

53e4e76

update_own_repo

c004301

tmp

4e22d6e

added yaw=0 current yaw

46ddb33

integrate existing uuv features

914742a

Merge branch 'hippocampus' of https://github.com/zarbokk/sitl_gazebo into uuv_extension

major renaming from uuv2 -> uuv_hippocampus

e7daa9d

minor bug fixes and clean up

c96786f

* added uuv_hippocampus model carrying a camera

9ed4ce0

* general code clean up + comments

added uuv_hippocampus_cam.world

9664823

minor changes to hippoc_cam world

89b3891

cleaning up

4a81dd4

minor changes to new uuv_hippocampus world

ae3af94

Merge branch 'uuv_extension' of https://github.com/danielduecker/sitl…

7b2020e

…_gazebo into uuv_extension

Merge branch 'uuv_extension' of https://github.com/danielduecker/sitl…

23f5fe6

…_gazebo into uuv_extension

Merge branch 'master' of https://github.com/px4/sitl_gazebo into uuv_…

22f8d9e

…extension

DanielDuecker requested review from dagar and Jaeyoung-Lim January 31, 2020 16:52

DanielDuecker mentioned this pull request Jan 31, 2020

add support for unmanned underwater vehicles: PX4/PX4-Autopilot#14079

Merged

remove >> frame='' << from pose in sdf files

a7f2623

Jaeyoung-Lim reviewed Feb 3, 2020

View reviewed changes

src/gazebo_uuv_plugin.cpp Outdated Show resolved Hide resolved

src/gazebo_uuv_plugin.cpp Outdated Show resolved Hide resolved

src/gazebo_uuv_plugin.cpp Outdated Show resolved Hide resolved

Jaeyoung-Lim reviewed Feb 3, 2020

View reviewed changes

models/uuv_hippocampus/uuv_hippocampus.sdf Outdated Show resolved Hide resolved

Jaeyoung-Lim reviewed Feb 3, 2020

View reviewed changes

DanielDuecker requested a review from Jaeyoung-Lim February 5, 2020 07:57

Jaeyoung-Lim approved these changes Feb 5, 2020

View reviewed changes

dagar merged commit 045156d into PX4:master Feb 5, 2020

Jaeyoung-Lim approved these changes Feb 5, 2020

View reviewed changes

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

underwater robot/uuv extension #402

underwater robot/uuv extension #402

DanielDuecker commented Jan 31, 2020

Jaeyoung-Lim left a comment •

edited

Loading

Jaeyoung-Lim Feb 3, 2020

DanielDuecker Feb 4, 2020

Jaeyoung-Lim Feb 4, 2020

DanielDuecker commented Feb 4, 2020 •

edited

Loading

Jaeyoung-Lim commented Feb 4, 2020

DanielDuecker commented Feb 4, 2020

Jaeyoung-Lim commented Feb 4, 2020

DanielDuecker commented Feb 5, 2020

Jaeyoung-Lim left a comment

underwater robot/uuv extension #402

underwater robot/uuv extension #402

Conversation

DanielDuecker commented Jan 31, 2020

Jaeyoung-Lim left a comment • edited Loading

Choose a reason for hiding this comment

Jaeyoung-Lim Feb 3, 2020

Choose a reason for hiding this comment

DanielDuecker Feb 4, 2020

Choose a reason for hiding this comment

Jaeyoung-Lim Feb 4, 2020

Choose a reason for hiding this comment

DanielDuecker commented Feb 4, 2020 • edited Loading

Jaeyoung-Lim commented Feb 4, 2020

DanielDuecker commented Feb 4, 2020

Jaeyoung-Lim commented Feb 4, 2020

DanielDuecker commented Feb 5, 2020

Jaeyoung-Lim left a comment

Choose a reason for hiding this comment

Jaeyoung-Lim left a comment •

edited

Loading

DanielDuecker commented Feb 4, 2020 •

edited

Loading