Lower-body Kinematics Estimation Using Inertial Measurement Units

Team 1

Overview

All codes implemented for this project can be found in the folder code while results can be found in the folder results.

All data can downloaded from the folders data_imu (containing IMU data) and data_mocap (containing marker-based motion capture data).

Implementation

All codes were implemented in Python 3.8 version. Below is the list of libraries needed to succesfully run the code:

• numpy
• math
• pandas
• matplotlib
• opensim (see https://simtk-confluence.stanford.edu:8443/display/OpenSim/Scripting to learn how to install the opensim library for constrained IMU inverse kinematics)
• plotly
• AHRS (see https://ahrs.readthedocs.io/en/latest/index.html to learn how to install the AHRS library for using sensor fusion algorithms)

Guideline

1. Use imu_unc_ik.py to obtain joint angles without using OpenSense (i.e., unconstrained IMU).
2. Use opensense_ik.py to obtain joint angles using OpenSense (i.e., constrained IMU).
3. Use xsens_to_opensim_all.py to convert orientaion estimated from sensor fusion algorithms (including Xsens filter) to the OpenSim format (i.e., .sto files).
4. Use comparison_sync.py to synchronize motion capture and IMU data.
5. Use comparison_unconstrained.py to compare unconstrained IMU-based kinematics (without OpenSense).
6. Use compare_opensim_opensense.py to compare constrained IMU-based kinematics (with OpenSense).

Unconstrained and constrained IMU inverse kinematics results of (1) and (2) will be stored in the folder results. (2) uses the .sto files generated by (3) and you can customize which .sto file is used for calibration and which is used for the actual calculations in (2). When running (4), (5), or (6), results from the folder results will be used to perform comparison between different methods.