alexanderdurr / goalgail

accompanying code for neurips submission "Goal-conditioned Imitation Learning"

Geek Repo:Geek Repo

Github PK Tool:Github PK Tool

Goal-conditioned Imitation Learning

Environment setup

conda env create -f environment.yaml

Run Experiment

The following command will run the three experiments as in Fig. 3 in the paper for both four rooms env and Fetch Pick&Place env.

Four rooms env: python sandbox/experiments/goals/maze/maze_her_gail.py

Point mass lego pusher env: python sandbox/experiments/goals/maze_lego/maze_lego_her_gail.py

Fetch Pick and Place env: python sandbox/experiments/goals/pick_n_place/pnp.py

Fetch StackTwo env: python sandbox/experiments/goals/pick_n_place/pnp_twoobj.py

Plot Learning Curves

The following command will reproduce the learning curves for two environments as in Fig. 3 in the paper.

Four rooms env: python plotting/gail_plot.py data/s3/fourroom fourroom

Point mass lego pusher env: python plotting/gail_plot.py data/s3/pointmass-block-pusher pointmass-block-pusher

Fetch Pick and Place env: python plotting/gail_plot.py data/s3/fetchpnp fetchpnp

Fetch StackTwo env: python plotting/gail_plot.py data/s3/stacktwo stacktwo

The generated figures can be found in folder figures.

About

accompanying code for neurips submission "Goal-conditioned Imitation Learning"


Languages

Language:Python 95.9%Language:HTML 1.5%Language:JavaScript 1.3%Language:Ruby 0.6%Language:CSS 0.4%Language:Mako 0.2%Language:Shell 0.2%