#  --------------------------------------------------------

This file repository contains supplementary data and code for the submission:
Fluidic Innervation Sensorizes Structures from a Single Build Material

#  --------------------------------------------------------

Ryan L. Truby, Lillian Chin, Annan Zhang, and Daniela Rus

MIT Computer Science and Artificial Intelligence Lab
Massachusetts Institute of Technology

#  --------------------------------------------------------

Contents:

- install
  -> Contains files to install the conda environment:
     - installation.txt: Installation instructions
     - environment.yml: Dependencies and required packages

- code
  -> Contains code for preprocessing, training, evaluation:
     - preprocessing.py: script to preprocess data
     - dataset.py: dataset class
     - models.py: model class for LSTM
     - utils.py: utility functions, including training and evaluating
     - main.py: script to train single model
     - main_hpsearch.py: script for hyperparameter search
     - evaluate.py: script to evaluate existing model on test data

- data
  -> Contains raw and preprocessed data
     - data.pkl: raw data files with all sequences in one array, separated by zero rows
     - data_all_list_rnd-seed1619.pkl: data separated into lists of training/validation/test sequences, statistics
     - data_train-val_seq-len64_val-ratio0.1_rnd-seed1619.pkl: training and validation data generated with sliding window method described in Materials and Methods
     - log_rnd-seed1619.txt: log file generated during preprocessing

- models
  -> Contains weights and log files of the trained models
     - search
       -> Contains results and models of hyperparameter search
     - retraining
       -> Contains models from final retraining with best architecture

- anim
  -> Contains code and files to generate renders of the sHSA platform:
     - best-model
       -> Contains: 
          - csv: ground truth and prediction files copied from evaluation of best model on test set
          - mp4: renders in 3D view, generated by animate.m
          - mp4_front-view: renders in front view, generated by animate.m
     - animate.m: MATLAB script to generate renders
     - calculate_*.m: Helper functions for animate.m


#  --------------------------------------------------------

Contact Annan Zhang (zhang@csail.mit.edu) for questions.