Generalized_MaxEnt_RL

This is the implementation for the paper of Generalized Maximum Entropy Reinforcement Learning (SSPG)

Set up env: (https://www.chenshiyu.top/blog/2019/06/19/Tutorial-Installation-and-Configuration-of-MuJoCo-Gym-Baselines/)

''' conda create -n mujoco-gym python=3.6 conda activate mujoco-gym '''

cd /locations_you_want ''' git clone https://github.com/openai/mujoco-py.git cd mujoco-py pip install -e . '''

install Gym (openai/mujoco-py#477)

Download MuJoCo and its licence from https://www.roboti.us/download.html

Extract amd move the file to the specified path (also add path to .bashrc)

Following the format: export LD_LIBRARY_PATH=/home/csy/.mujoco/mujoco200/bin${LD_LIBRARY_PATH:+:${LD_LIBRARY_PATH}}

''' pip install gym[all]==0.15.3

May needs: sudo apt-get install libosmesa6-dev (ethz-asl/reinmav-gym#35) '''

Install Pytorch:

''' conda install -c pytorch pytorch

'''

Test env

''' python gym_test.py '''

MuJoCo is deterministic env

Create testing env (Grid)