d-reflex

D-reflex experiments

python

python scripts and dataset

dataset

situations dataset of 2000 samples used in the revised version of the paper (with damage combinations and reject everything but the hand contact)

scripts

written using JupyterLab and Jupytext.

t22_04_20(NN_training).py

Jupytext script to train and evaluate a neural network using the collected dataset.

22_05_05(reaction_delay).py

Jupytext script to evaluate the effect of the delay before activating the reflex

22_05_06(analyse_dataset).ipynb

jupyter notebook to analyse the origin of the unavoidable/avoidable using a decision tree

22_05_07(generate_data).py

Jupytext script to colect data using the c++ WBC and Dart simulation.

22_05_09(prepare_data_for_NN).ipynb

jupyter script to clean the dataset for the NN training

22_05_16(RAL_Revision_for_paper_video).py

Jupytext script to generate the simulation videos for the paper

22_05_16(RAL_revision_friction).py

Jupytext script to evaluate the effect of the friction

22_05_16(RAL_revision_main_figure).py

Jupytext script to generate the snapshots for the main figure

cpp

talos

urdf files used for the experiments.

d-reflex

custom tasks, behaviors and controller for D-Reflex.

inria_wbc

Whole-Body Controller for the TALOS.