RL Audio

Codebase for audio communication reinforcement learning (RL) bandit algorithm. This repository enables a virtual robot to learn to communicate different functional robot states using parameterized audio.

Link to Research Paper

TBA

Main Files

This method was deployed and tested in a user study. This study took the form of a Jupyter notebook. We provide the original interactive survey (Jupyter Notebook) with added step-by-step markdown comments describing the study procedure to enable reproducibility.

The Jupyter notebook used for our study is here:

https://github.com/liamreneroy/RL_audio/blob/main/notebooks/study_notebook.ipynb

The sound libraries used for each robot are here:

https://github.com/liamreneroy/RL_audio/tree/main/notebooks/audio

The arrays and images for Informed Initializations are here:

https://github.com/liamreneroy/RL_audio/tree/main/notebooks/arrays

The statistical analyses for our three hypotheses (H1, H2, H3) are here:

https://github.com/liamreneroy/RL_audio/blob/main/stats/

All the raw data collected throughout our study is here:

https://github.com/liamreneroy/RL_audio/blob/main/notebooks/user_data/response_book.xlsx

Packages to Install:

pygame (see this webpage ~ https://www.pygame.org/wiki/GettingStarted)
jupyterlab, numpy, termcolor, openpyxl, nbconvert-webpdf

Either use:

sudo apt-get install <package_name>
python3 -m pip install <package_name>
conda install -c conda-forge <package_name>

Example using conda:

conda install -c conda-forge <package_name>

• jupyterlab or notebook
• numpy
• termcolor
• openpyxl
• nbconvert-webpdf

Owner:

Liam Rene Roy [email protected]