README.md

mc_opensim

Tools for running mc_reconstruction outputs through OpenSim's IK solver

A set of scripts to:

• Fill gaps and smooth the output of a .c3d from mc_reconstruction.
• Batch convert .c3d data from mc_reconstruction to an OpenSim compatible format.
• Batch process converted outputs from mc_reconstruction outputs through OpenSim's IK solver.

Note:

./isbs21_scripts contains analysis scripts to produce results and figures for ISBS 2021 abstract.

./paper2_scripts contatins analysis scropts to produce results and figures for BioCV Paper 2.

Installation

1. Get hold of and compile OpenSim. OpenSim provide an install script - we have included a modified copy of the install script in mc_opensim/opensim-core-linux-build-script.sh which you may find convenent. The modifications allow you to build and install OpenSim somewhere other than you're home directory. After OpenSim has build, remember to install the python modules. For example:

bash opensim-core-linux-build-script.sh -p /opt/opensim
cd /opt/opensim/install/sdk/Python
sudo python setup.py install

You can then test with python3 -m opensim

2. Once OpenSim is playing ball, clone the repository from camera-mc-dev:

foo@bar:~$ git clone ...

foo@bar:~$ cd mc_opensim

foo@bar:~./mc_opensim$ python3 -m venv ./venv --system-site-packages

foo@bar:~./mc_opensim$ source venv/bin/activate

(venv) foo@bar:~./mc_opensim$ python3 -m pip install update pip
(venv) foo@bar:~./mc_opensim$ python3 -m pip install -r requirements.txt

Usage

FillSmoothC3D.py - Use to fill gaps and smooth a .c3d file.

• You can use this in one of two ways:
  1. Simply pass a list of .c3d files to the script on the command line, as well as the noise parameters for the Kalman filter.
  2. Set the PATH variable in the config.py file to the path of the session directory.
     • the script will search beneath that dir for any .c3d files and try to smooth them.
     • noise parameters for the Kalman smoother will be taken from the config file.
     • If one or both of the noise parameters are set negative, then the script will ask PyKalman to estimate suitable parameters.
     • We use a constant acceleration filter.
     • increasing the transition noise or the reducing the observation noise allows the state to deviate from that assumption
     • reducing the transition noise or increasing the observation nose encourages the state to stick to that assumption

(venv) foo@bar:~./mc_opensim$ python3 FillSmoothC3D.py False 0.01 15 file00.c3d file01.c3d ... fileNN.c3d

(venv) foo@bar:~./mc_opensim$ python3 FillSmoothC3D.py True

TrcGenerator.py - Use to convert .c3d files to OpenSim .trc format.

• Update config.py by:
  • Setting PATH variable to the absolute path of the session directory.
    • The script will walk through the filesystem looking for .c3d files and converting them.
  • Setting other trc settings as desired (details of each setting given in config file).
• Run:

(venv) foo@bar:~./mc_opensim$ python3 TrcGenerator.py

BatchIK.py - Use to generate scaled osim model and run IK solver on motion files

• Ensure that .trc files have been generated.
• Update config.py by:
  • Setting PATH variable to the absolute path of the session directory.
  • Setting other scaling and IK settings as desired (details of each setting given in config file).
  • For fine-grained control of IK settings see .xml files in ./configs/.
• Run:

(venv) foo@bar:~./mc_opensim$ python3 BatchIK.py

Developer Notes

When running the IK solver, we currently assume the following:

• For markerbased data, we assume that the scaled model (.osim) is located in the same directory as the .trc files. e.g.

session_dir
│
├──trial_01.trc
├──trial_02.trc
└──scaled_static.osim

• For markerless data, we assume that the scaled model (.osim) is located in the same parent directory as the motion trials e.g.

session_dir
│
└──trial_01
|   |
|   └──trial.trc
└──trial_02
|   |
|   └──trial.trc
└──STATIC_01
    |
    └──scaled_static.osim