search

robotology / icub-models-generator

Resources and programs to generated models (URDF, SDF) of the iCub robot
14 stars 23 forks source link

Add model for iCubGenova11 #252

Closed traversaro closed 8 months ago

traversaro commented 8 months ago

The iCubGenova11 is an iCub 2.7, so we generated the model accordingly.

It also fixes https://github.com/robotology/icub-models-generator/issues/255 to ensure that CI is green. I manually run the CI to push the generated files in https://github.com/robotology/icub-models/tree/addicubgenova11 for testing @martinaxgloria .

traversaro commented 8 months ago

This is required to use the "modern" wholebodydynamics device on iCubGenova11 . fyi @martinaxgloria

traversaro commented 8 months ago

I manually run the CI to push the generated files in https://github.com/robotology/icub-models/tree/addicubgenova11 for testing @martinaxgloria .

The CI failed with:

fatal: Remote branch addicubgenova11 not found in upstream origin

I forgot that I add to create manually the branch in icub-models, we should document this somewhere. I relaunched the CI in https://github.com/robotology/icub-models-generator/actions/runs/7461748396 .

martinaxgloria commented 8 months ago

This is required to use the "modern" wholebodydynamics device on iCubGenova11 . fyi @martinaxgloria

Thank you very much @traversaro!

traversaro commented 8 months ago

It took some back and forth, but now the test branch has the commit that adds iCubGenova11, see https://github.com/robotology/icub-models/commit/bcbae74de89fa996dd45e1cc593866b0eced6464 . Probably some change in pip dependencies also did some spurious changes to the files, but the core change is the new file for iCubGenova11.

traversaro commented 8 months ago

@martinaxgloria @Nicogene feel free to approve if you tested this!

martinaxgloria commented 8 months ago

@martinaxgloria @Nicogene feel free to approve if you tested this!

Hi @traversaro, I'm going to test it today. Thanks!

martinaxgloria commented 8 months ago

I was thinking, since it is the model for iCubGenova11, which has the legs disabled (and also the FT on them), could we exclude them also in the urdf? Or they will be enabled again in the future on the real robot?

Or, better, there's a way with wholeBodyDynamics to choose which FT sensors are used (for example with a parameter)? In this way, there's no need to exclude the FT of the legs and the feet from the URDF

cc @Nicogene @traversaro

martinaxgloria commented 8 months ago

According to this issue the only workaround so far is to modify the URDF. Then, I think we could disable at least the FT sensors of the lower body

Nicogene commented 8 months ago

I was thinking, since it is the model for iCubGenova11, which has the legs disabled (and also the FT on them), could we exclude them also in the urdf? Or they will be enabled again in the future on the real robot?

Good point, since the legs are electrically disabled I would remove them, it will fall if used in gazebo but maybe we make it just fixed?

traversaro commented 8 months ago

I was thinking, since it is the model for iCubGenova11, which has the legs disabled (and also the FT on them), could we exclude them also in the urdf? Or they will be enabled again in the future on the real robot?

Good point, since the legs are electrically disabled I would remove them, it will fall if used in gazebo but maybe we make it just fixed?

We are not generating any Gazebo-loadable model (i.e. with a model.config) out of this model, so I think we should be safe on this.

traversaro commented 8 months ago

According to this issue the only workaround so far is to modify the URDF. Then, I think we could disable at least the FT sensors of the lower body

Good point. Could we think to add a Python script or similar that does this to the model generated by simmechanics (and eventually by creo2urdf)? I guess it should be a task that can be easily done with the help of ChatGPT or similar LLM tech.

traversaro commented 8 months ago

According to this issue the only workaround so far is to modify the URDF. Then, I think we could disable at least the FT sensors of the lower body

Good point. Could we think to add a Python script or similar that does this to the model generated by simmechanics (and eventually by creo2urdf)? I guess it should be a task that can be easily done with the help of ChatGPT or similar LLM tech.

See https://chat.openai.com/share/69784828-87f9-4a53-b805-55d02d66384b for a starting point (I do not have time to work on this further), not sure if it works. Once we have that, I am happy to integrate the resulting Python script in CMake.

Nicogene commented 8 months ago

According to this issue the only workaround so far is to modify the URDF. Then, I think we could disable at least the FT sensors of the lower body

Good point. Could we think to add a Python script or similar that does this to the model generated by simmechanics (and eventually by creo2urdf)? I guess it should be a task that can be easily done with the help of ChatGPT or similar LLM tech.

Ok for me!

traversaro commented 8 months ago

@martinaxgloria @Nicogene Probably we can start merging this, and leave the disabling of the FT sensors in the future PR? In any case, this is for sure a net improvement over not having a model at all.

Nicogene commented 8 months ago

@martinaxgloria @Nicogene Probably we can start merging this, and leave the disabling of the FT sensors in the future PR? In any case, this is for sure a net improvement over not having a model at all.

Ok for me!

martinaxgloria commented 8 months ago

Ok for me :+1:

traversaro commented 8 months ago

I opened https://github.com/robotology/icub-models/issues/229 for the follow up.