Code contained within the subfolder goal_conditioned_baselines is taken from https://github.com/knowledgetechnologyuhh/goal_conditioned_RL_baselines, with some modifications. The main modifications are to be able to run the episodes in parallel using multiprocessing, though I have also made some other small changes to get it to work with my environment and improve performance. The two mujoco xml files in mujoco_sim are also modified from that repo to include the heightmap data. anymal-environment uses the ANYmal model (also modified) from https://github.com/google-deepmind/mujoco_menagerie and the files contained in the assets folder are from there.
Install all necessary requirements (see requirements.txt)
To be able to visualise the robot agent, you'll also want to install the mujoco-python-viewer package
pip install mujoco-python-viewer
and update line 8 on mujoco_sim/mujoco_simulation.py
To train an ant agent, run
python3 train_agent.py <savepath>
replacing "" with the path you want to save the agent to.
To run an agent, run
python3 testrun_agent.py <loadpath>
replacing "" with the path to the previously saved agent.