Organization

The repository is organized into 8 different folders:

• analyses
  • Jupyter notebook containing code used to analyze data
• cozmo (files organized and named by timestamp (ID) of behavior; not all functions will have state information and vice versa)
  • Function folder containing text files listing the functions that were ran for a behavior
  • States folder containing json files listing the internal state changes of Cozmo while behavior was running
• misty (files named by timestamp (ID) of behavior; not all functions will have state information and vice versa)
  • Function folder containing text files listing the functions that were ran for a behavior
  • States folder containing json files listing the internal state changes of Misty while behavior was running
• cozmo_Filtered (files named by timestamp (ID) of behavior; all functions will have state information and vice versa)
  • Function folder containing text files listing the functions that were ran for a behavior
  • States folder containing json files listing the internal state changes of Cozmo while behavior was running
• misty_Filtered (files named by timestamp (ID) of behavior; all functions will have state information and vice versa)
  • Function folder containing text files listing the functions that were ran for a behavior
  • States folder containing json files listing the internal state changes of Misty while behavior was running
• raw_data
  • cozmo_functions.pkl and misty_functions.pkl
    • pkl files containing dictionary that maps behavior timestamps to functions used to run behavior for Cozmo and Misty
  • cozmo_filtered_data.pkl and misty_filtered_data.pkl
    • pkl files containing dataframes of cleaned data for Cozmo and Misty
  • cozmo_states.pkl and misty_states.pkl
    • pkl files containing dataframes of internal state data for Cozmo and Misty
  • functions_cozmo.txt and functions_misty.txt
    • text files containing all functions of every behavior ran labeled by behavior's timestamps
  • response_cozmo copy.csv and response_misty copy.csv
    • csv file containing descriptions and affect labels for each behavior labled by behavior's timestamps
• resources
  • cozmo_faces folder containing images used to generate Cozmo's faces
  • misty_faces_black folder containing images used to generate black Misty's faces
  • misty_faces_white folder containing images used to generate white Misty's faces
• src
  • extract_cozmo_states.py
    • runs all Cozmo behaviors and extract internal state data
  • extract_misty_states.py
    • runs all Misty behaviors and extract internal state data
  • main.py
    • driver code used to run extract_misty_states.py
  • novikova.py
    • maps internal robot states to novikova features
  • MapFunctionsToInternalStates.ipynb
    • used to create cozmo_functions.pkl and misty_functions.pkl files organizing behavior timestamps to array of functions
  • semdial2020_model.ipynb
    • model used to map internal states to emotions
  • zero-shot.ipynb
    • BART model used for zero-shot learning task mapping descriptions to emotions