An environment for kinship-aligned multi-agent reinforcement learning — where agents with diverging interests either learn to cooperate or see their lineage go extinct.
The best place to understand why Territories and follow the experimental results is the through the blog series: Emerging a Society with MARL.
The best place to understand RL environments written in C is by studying PufferLib.
Install raylib.h and the requirements.txt. Compile the environment using sh build.sh.
The you can use the environment in Python. See src/territories.py for an example.