This project explores interactive and collaborative evolution of neural network controllers for Minecraft agents. Interactive means that it will be the users that guide evolution, and collaborative that they can save their controllers and continue evolution where others left off.
Please see the developer manual for more information about the code and how to use and modify it.
CrowdAI was a postdoc project by Pablo González de Prado Salas at IT University of Copenhagen, supervised by Sebastian Risi. It builds on our own work on modular neural networks (Github repository here), which itself is an adaptation of SharpNeat. We use Project Malmo to interact with Minecraft.
Please refer to the License file for more details.