Body Brain Behavior - Quality Diversity

Install guide

Clone the repo and its submodules and cd into it

git clone --recurse-submodules https://github.com/giorgia-nadizar/BBB-QD.git
cd BBB-QD

If you have already cloned the repository without its submodules, run the following command to fix it

git submodule update --init --recursive

Normal usage

Install some required libraries

sudo apt-get install xorg-dev libglu1-mesa-dev
sudo apt-get install cmake
sudo apt-get install freeglut3-dev
sudo apt-get install libglfw3
sudo apt-get install libgl1-mesa-glx

Create conda environment and install a required library

conda env create -f environment.yml
conda activate bbbqd
conda install -c conda-forge libstdcxx-ng -y

Build and install package

python setup.py install

Docker

The Docker file is bbbqd.Dockerfile To execute the project with docker build the file first

docker build --target run-image -t bbb-qd:0.0.1 -f bbbqd.Dockerfile .

and the run it

docker run --gpus all -it -v "$(pwd)/results:/app/results" bbb-qd:0.0.1

the flag --gpus is needed to allow the usage of the GPU, if available; the flag -v enables the mounting on the container of the folder results. They can both be removed if needed.

To run the container in interactive mode add the flag --entrypoint bash.

Testing installation

To test your installation was successful you can run examples/install_tester.py which will test that both jax and evogym are properly installed. This script will print a bunch of arrays.

Otherwise, you can run examples/viz_tester.py, which will launch an evogym simulation and its visualization.

Experiments, results, videos

The folder experiments contains the code for running all the experiments performed so far. Namely, cgp_ga.py evolves a CGP controller with both global and modular controllers for 4 fixed bodies. Instead, cgp_body_ga.py evolves the body-within a 5x5 grid-and the CGP controller for both global and modular controllers.

The analysis of results is in the folder analysis, in the preliminary.ipynb notebook.

Some videos of evolved robots are available in the videos folder.