aikkala / o2mconverter Goto Github PK

Hi,

I just came across this project. It's very helpful for me. Thanks for sharing.

But I got a problem with the gait models (FullBodyModel_SimpleArms_Hamner2010_Markers_v2_0, gait10dof18, gait2392, leg6dof9). The shank segment and the thigh segment overlap each other when I load the converted xml model to mujoco (I'm using mujoco 2.1). Do you also have the same issue?

Thanks.

Great works! But is it possible to create MJCF motion file from the opensim motion (.mot ) ?

trying to create the environment from yml file gives the following error:

(base) D:>conda env create --name O2MConverter --file=conda_env.yml
Retrieving notices: ...working... done
Collecting package metadata (repodata.json): done
Solving environment: failed

ResolvePackageNotFound:

• _libgcc_mutex==0.1=conda_forge
• xorg-libxfixes==5.0.3=h516909a_1004
• krb5==1.14.6=0
• lz4-c==1.8.3=he1b5a44_1001
• libxcb==1.13=h14c3975_1002
• xorg-libxext==1.3.4=h516909a_0
• freetype==2.10.1=he06d7ca_0
• patchelf==0.10=he1b5a44_0
• vtk==8.2.0=py36ha8e561a_202
• libtiff==4.1.0=hc7e4089_6
• xorg-fixesproto==5.0=h14c3975_1002
• xorg-xproto==7.0.31=h14c3975_1007
• icu==64.2=he1b5a44_1
• openblas==0.3.9=he1df0ab_0
• zlib==1.2.11=h516909a_1006
• pandas==1.0.3=py36h830a2c2_0
• xorg-libsm==1.2.3=h84519dc_1000
• zstd==1.4.4=h3b9ef0a_2
• scikit-learn==0.22.2.post1=py36hcdab131_0
• libstdcxx-ng==9.2.0=hdf63c60_2
• libuuid==2.32.1=h14c3975_1000
• libgcc-ng==9.2.0=h24d8f2e_2
• tk==8.5.19=2
• sqlite==3.13.0=1
• bzip2==1.0.8=h516909a_2
• tbb==2020.1=hc9558a2_0
• libwebp-base==1.1.0=h516909a_3
• xorg-xextproto==7.3.0=h14c3975_1002
• freeglut==3.0.0=hf484d3e_1005
• pthread-stubs==0.4=h14c3975_1001
• xorg-libxi==1.7.10=h516909a_0
• libssh2==1.8.0=h1ad7b7a_1003
• libnetcdf==4.6.2=hbdf4f91_1001
• readline==6.2=0
• xorg-libx11==1.6.9=h516909a_0
• xorg-libxt==1.2.0=h516909a_0
• xorg-libxau==1.0.9=h14c3975_0
• hdf4==4.2.13=hf30be14_1003
• xorg-inputproto==2.3.2=h14c3975_1002
• libpng==1.6.37=hed695b0_1
• xz==5.2.4=h516909a_1002
• jpeg==9c=h14c3975_1001
• curl==7.61.0=h93b3f91_2
• _openmp_mutex==4.5=1_llvm
• libopenblas==0.3.9=h5ec1e0e_0
• ncurses==5.9=10
• scipy==1.4.1=py36h2d22cac_2
• xorg-kbproto==1.0.7=h14c3975_1002
• xorg-libice==1.0.10=h516909a_0
• hdf5==1.10.4=nompi_h3c11f04_1106
• openssl==1.0.2u=h516909a_0
• libxml2==2.9.10=hee79883_0
• llvm-openmp==9.0.1=hc9558a2_2
• jsoncpp==1.8.4=hc9558a2_1002
• expat==2.2.9=he1b5a44_2
• libiconv==1.15=h516909a_1006
• libgfortran-ng==7.3.0=hdf63c60_5
• xorg-libxdmcp==1.1.3=h516909a_0

Hi,

great project!

I am trying to run the scripts to reproduce you results but the function estimate_joint_error function is missing in Utils.py

Thanks for the open source converter. I was following your steps as per Readme and when it reaches the calculate joint errors in process test runs, the format of data is different from your pickle data dump from optimize mujoco parameters.

From process_test_runs.py calculate_joint_errors_function

states = data[run_idx]["states"]
controls = data[run_idx]["controls"]
timestep = data[run_idx]["timestep"]
run = data[run_idx]["run"]

From optimize_mujoco_parameters.py collect_data_from_runs function

data.append({"qpos": qpos_values, "qvel": qvel_values, "controls": controls.values,
                     "state_names": state_names, "muscle_names": list(controls),
                     "timestep": env.timestep, "success": success, "run": run, "num_evals": num_evals})

Hello,

Thanks for the handy package! I'm trying to convert the HYOID model of the head and neck, however the BushingForce of OpenSim is not included. The BushingForce in this case is the rotational stiffness and damping in the joints between the cervicals.

Since I'm new to MuJoCo, my question is how to include this in the MuJoCo XML, is it just the 'stiffness' and 'damping' parameters in the hinge joints of those bodies (cervicals)?

I think the conda environment doesn't need all those packages and it breaks

Solving environment: failed

ResolvePackageNotFound: 
  - pandas==1.0.3=py36h830a2c2_0
  - openssl==1.0.2u=h516909a_0
  - bzip2==1.0.8=h516909a_2
  - _openmp_mutex==4.5=1_llvm
  - libssh2==1.8.0=h1ad7b7a_1003
  - patchelf==0.10=he1b5a44_0
  - xorg-libsm==1.2.3=h84519dc_1000
  - scipy==1.4.1=py36h2d22cac_2
  - libxml2==2.9.10=hee79883_0
  - xorg-libxdmcp==1.1.3=h516909a_0
  - expat==2.2.9=he1b5a44_2
  - xorg-xextproto==7.3.0=h14c3975_1002
  - libpng==1.6.37=hed695b0_1
  - libxcb==1.13=h14c3975_1002
  - freetype==2.10.1=he06d7ca_0
  - icu==64.2=he1b5a44_1
  - freeglut==3.0.0=hf484d3e_1005
  - xorg-libx11==1.6.9=h516909a_0
  - pthread-stubs==0.4=h14c3975_1001
  - xorg-libxfixes==5.0.3=h516909a_1004
  - scikit-learn==0.22.2.post1=py36hcdab131_0
  - libwebp-base==1.1.0=h516909a_3
  - xorg-inputproto==2.3.2=h14c3975_1002
  - xorg-xproto==7.0.31=h14c3975_1007
  - jpeg==9c=h14c3975_1001
  - openblas==0.3.9=he1df0ab_0
  - jsoncpp==1.8.4=hc9558a2_1002
  - tbb==2020.1=hc9558a2_0
  - zstd==1.4.4=h3b9ef0a_2
  - libopenblas==0.3.9=h5ec1e0e_0
  - xorg-kbproto==1.0.7=h14c3975_1002
  - xorg-libxt==1.2.0=h516909a_0
  - vtk==8.2.0=py36ha8e561a_202
  - libgcc-ng==9.2.0=h24d8f2e_2
  - lz4-c==1.8.3=he1b5a44_1001
  - libuuid==2.32.1=h14c3975_1000
  - hdf5==1.10.4=nompi_h3c11f04_1106
  - xz==5.2.4=h516909a_1002
  - libtiff==4.1.0=hc7e4089_6
  - libstdcxx-ng==9.2.0=hdf63c60_2
  - _libgcc_mutex==0.1=conda_forge
  - xorg-fixesproto==5.0=h14c3975_1002
  - xorg-libxi==1.7.10=h516909a_0
  - libgfortran-ng==7.3.0=hdf63c60_5
  - llvm-openmp==9.0.1=hc9558a2_2

I believe only those are needed

channels:
  - kidzik
  - conda-forge
  - https://conda.anaconda.org/conda-forge/
  - defaults
dependencies:
  - vtk=8.1.1   
  - xmltodict=0.12.0
  - pip:
    - matplotlib==3.2.1
    - mujoco-py==2.0.2.9
    - numpy==1.18.2
    - pyquaternion
    - sklearn 
    - pandas
    - scikit-video