TAXONS

Task Agnostic eXploration of Outcome spaces through Novelty and Surprise.

This is the code of the paper: Unsupervised Learning and Exploration of Reachable Outcome Space

---

To install run:

pipenv shell --three
python setup.py install

We also provide a containerized version running in Singularity at: https://github.com/GPaolo/taxons_sif

---

Dependencies

NB: if you're using the virtualenv, activate it before installing the dependencies.

Pybullet Gym

I am using a slightly modified version of pybulletgym than the original found here: https://github.com/benelot/pybullet-gym.

To install it, activate the virtual env, go in the external folder and run:

git clone https://github.com/GPaolo/pybullet-gym.git
cd pybullet-gym
pip install -e .

If you want more informations, look at the README there.

Fastsim

TAXONS needs Fastsim to be installed. I am using a slightly modified version of the original.

To install it we first need to download Pyfastsim:

git clone https://github.com/alexendy/pyfastsim

Now we need to install libfastsim in pyfastsim. To do so, execute:

cd external/pyfastsim
git clone https://github.com/GPaolo/libfastsim.git
cd libfastsim
git checkout patch-1
git pull origin patch-1
patch -p1 < ../fastsim-boost2std-fixdisplay.patch
./waf configure
./waf build
./waf install

NB If it complains that cannot find boost, then install it by running:

sudo apt-get install libboost-all-dev

Now we can install pyfastsim by running, in the external/pyfastsim folder:

python setup.py install

Finally we can install fastsim-gym. To do so, activate the virtual env and enter the external folder. Then do:

cd external
git clone https://github.com/GPaolo/fastsim_gym.git
git checkout patch-1
git pull origin patch-1
python setup.py install

---

The file .env will be loaded automatically with pipenv shell or pipenv run your_command and the environment variables will be available.

NB: within Pycharm you need the plugin Env File to load it automatically (access Env File tab from the Run/Debug configurations). You will have to run PyCharm from the shell itself from inside the activated virtualenv

Running

To run the algorithm you just need to launch:

python scripy/train.py

If you want to change the experiment parameters, go to: script/parameters.py

To plot the results, just run:

python scripts/plot.py