This is a repository of jupyter notebooks and python scripts used for analysing in the paper on estimating smoothness from IMU data.
The code was developed for executing it in Google Colab. So, to test this code, you will need to download the repository, and upload it into your Google Drive. Once uploaded, you should be able to run the Python notebooks files using Google Colab. The data generated from the notebooks are saved on Google Drive in the data folder of the respository.
There is a variable that needs to be modified in each notebook when you want to execute the code from your Google Drive. This is explained the individual notebooks.
There are four notebooks in this repositoy:
GenerateMJTForVirtualIMU.ipynb
Generates the simulated MJT movements with different via-points using CVXPY.
Demo notebook that show how CVXPY is used to generate MJT movements for a given set of iva-points.
Generates the plots demonstrating the properties of the LDLJ-A measure. This notebook requires the simulated data generated by GenerateMJTForVirtualIMU.ipynb
ReconstructedIMUMovements.ipynb
Generates simulated IMU data to evaluate the effect of reconstruction error on the LDLJ-A on translationnal movement, and SPARC and LDLJ-V on gyroscope data. This notebook also requires the simulated data generated by GenerateMJTForVirtualIMU.ipynb
If you are interested in the data used in the paper, then you can download the entire folder (4-5GB) containing the code and data from here.
Contact the author, Sivakumar Balasubramanian at siva82kb@gmail.com