/Reinforcement-Learning-with-Adaptive-Curriculum-Dynamics-Randomization-for-Fault-Tolerant-Robot-Cont

https://arxiv.org/abs/2111.10005

Primary LanguagePython

Adaptive Curriculum Dynamics Randomization

arXiv

ant

Set up

conda create -n acdr python=3.8
conda activate acdr
pip install -r requirements.txt

Run

# How to run ACDR_easy2hard algorithm
python run.py --train --agent_id=baseline --seed=1 --algo=CDR-v1

# How to run ACDR_hard2easy algorithm
python run.py --train --agent_id=baseline --seed=1 --algo=CDR-v2