CODAS

The Official PyTorch Code for "Cross-Modal Domain Adaptation for Cost-Efficient Visual Reinforcement Learning"

Tensorflow-version code: https://github.com/xionghuichen/CODAS

Code Structure

CODAS

|- models: code for models of CODAS

|- data: the precollect dataset, pre-trained dynamics model, environments are saved here

|- reset_able_mj_env: environment related code for CODAS

|- configs: task configurations

|- scripts: scripts to run codas

    |- run_data_collect.py: script to collect data of MuJoCo in the target domain

    |- train.py: script to train codas

Quick Start

# install python environment for CODAS
git clone --recursive https://github.com/jiangsy/mj_envs
git clone https://github.com/xionghuichen/CODAS
git clone https://github.com/jiangsy/mjrl
cd ../mj_envs/
pip install -e .
cd ../mjrl
pip install -e .
cd ../CODAS
pip install -e .

# the working directory is ./scripts
cd scripts

# run data collection in the target domain
python run_data_collect.py --env_id {task name} # to run data collect in hand DAPG envs, use the run_data_collect_robot.py script
# train codas
python train.py --env_id {task_name}

The training logs can be found in {your CODAS path}/log.