Modify ci to install and test RobotDynamicsEstimator library

[isorrentino]

Add bayes-filters-lib dependency (https://github.com/robotology/bayes-filters-lib/tree/master) to the CI to install the RobotDynamicsEstimator library.
PS. I do not know if I missed something as it is the first time I have modified the CI.
cc @GiulioRomualdi

[GiulioRomualdi]

Can you add it also in the conda ci?

[GiulioRomualdi]

Moreover, I think there is a typo /bug in the ci file you modified since the ci didn't start

[isorrentino]

Moreover, I think there is a typo /bug in the ci file you modified since the ci didn't start

I just fixed it.

[isorrentino]

Can you add it also in the conda ci?

The package (bayes-filters-lib) needed to install RobotDynamicsEstimator is not available in conda (see https://anaconda.org/robotology/repo). What do you suggest to install it?

[GiulioRomualdi]

Hi @isorrentino it's in conda-force https://anaconda.org/conda-forge/libbayes-filters-lib

conda install -c conda-forge libbayes-filters-lib

[GiulioRomualdi]

I made changes to the cache used in the CI at this URL: a52d28e. These changes were necessary to ensure the compilation of the 'bayes-filters-lib.' Additionally, I included 'bayes-filters-lib' in the list of libraries for build verification, in case the library is not enabled.

[GiulioRomualdi]

Waiting for a green CI

[GiulioRomualdi]

There seems there is an error in bipedal-locomotion-framework/src/Estimators/src/AccelerometerMeasurementDynamics.cpp:192:89

[isorrentino]

Hi @GiulioRomualdi, it seems it is failing only on Windows now, but I do not understand why.

[GiulioRomualdi]

On Windows the ci fails with https://github.com/ami-iit/bipedal-locomotion-framework/actions/runs/6682235198/job/18157187111?pr=746#step:13:6564

[GiulioRomualdi]

It fails in finding distutils. I think this is normal since windows ships python 3.12 and as explained here distutils is deprecated and nolonger shipped with a vanilla installation of python 3.12

[GiulioRomualdi]

Here is the line causing the ci failure

https://github.com/robotology/icub-models/blob/master/python/CMakeLists.txt#L17

A possibility is to use directly sysconfig as in https://docs.python.org/3/library/sysconfig.html

Perhaps @traversaro has some idea how to change that command

[traversaro]

Yes, see robotology/robotology-superbuild#1238 .

[traversaro]

Once we merge:

• Fix code to obtain Python relative install dir in Python 3.12 robotology/icub-models#219
• CMake: Permit to explictly specify Python installation directory robotology/icub-models#218

I think I would do a new icub-models tag.

[GiulioRomualdi]

Updating the version of icub-models macOS/ubuntu is not able to find the model anymore :(

https://github.com/ami-iit/bipedal-locomotion-framework/actions/runs/6690923073/job/18177160110?pr=746#step:24:69

[GiulioRomualdi]

Here is the issue: https://github.com/search?q=repo%3Aami-iit%2Fbipedal-locomotion-framework%20iCubGazeboV2_5_plus&type=code

the new version of icub-models https://github.com/robotology/icub-models/releases/tag/v2.4.0 rename iCubGazeboV2_5_plus to iCubGazeboV2_5_KIT_007

[traversaro]

Here is the issue: https://github.com/search?q=repo%3Aami-iit%2Fbipedal-locomotion-framework%20iCubGazeboV2_5_plus&type=code

Wow, Hyrum's Law at its finest. There is any reason for using the model with the ankle with the changed joint limits (as described in https://icub-tech-iit.github.io/documentation/upgrade_kits/ankle_for_stairs/support/)?

[GiulioRomualdi]

That code was written by @prashanthr05 I don't know why he decided to use that model. I can try to run the test with iCubGazeboV2_5

[GiulioRomualdi]

I got the following error

/home/gromualdi/robot-code/robotology-superbuild/build/install/share/iCub/robots/iCubGazeboV2_5/model.urdf
[2023-10-30 12:07:11.639] [thread: 27310] [blf] [debug] [StdImplementation::getParameterPrivate] Parameter named 'use_model_info' not found.
[2023-10-30 12:07:11.639] [thread: 27310] [blf] [warning] [FloatingBaseEstimator::initialize] The parameter handler could not find " use_model_info " in the configuration file.Setting to default value true
[ERROR] Model :: getFrameIndex : Frame named root_link_imu_acc not found in the model.
[2023-10-30 12:07:11.639] [thread: 27310] [blf] [error] [FloatingBaseEstimator::ModelComputations::setBaseLinkAndIMU] Specified IMU frame not available in the loaded URDF Model.
[2023-10-30 12:07:11.639] [thread: 27310] [blf] [error] [FloatingBaseEstimator::initialize] Could not load model related parameters.

I'm afraid that root_link_imu_acc is not defined in iCubGazeboV2_5

Still I can use iCubGazeboV2_7
https://github.com/robotology/icub-models/blob/d8919f77bb57462ea05c992999fa59582ae3941e/iCub/robots/iCubGazeboV2_7/model.urdf#L2555

[GiulioRomualdi]

Unfortunatelly I cannot use iCubGazeboV2_7 since root_link_imu_acc is only a sensor there while iCubGazeboV2_5_KIT_007 contains also the link. See here: https://github.com/search?q=repo%3Arobotology%2Ficub-models%20root_link_imu_acc&type=code

[GiulioRomualdi]

Still iCubGenova02 is fine!

[traversaro]

Unfortunatelly I cannot use iCubGazeboV2_7 since root_link_imu_acc is only a sensor there while iCubGazeboV2_5_KIT_007 contains also the link. See here: https://github.com/search?q=repo%3Arobotology%2Ficub-models%20root_link_imu_acc&type=code

Thanks for reporting this, it is a bug: robotology/icub-models#223 .

[prashanthr05]

That code was written by @prashanthr05 I don't know why he decided to use that model. I can try to run the test with iCubGazeboV2_5

Yes, you already figured it out 🚀 , it was for the IMU on the root link. I should have documented it to save you some debugging time.