Village AI War

A Gymnasium environment for a small RTS-style village game. You can watch a quick demo, play against a trained MAPPO bot team, or train your own policies.

Training in this repo is MAPPO only (multi-agent PPO with a shared critic). Everything else is the game simulation, rendering, and helpers.

---

Install and first run

cd VillageAI_War
python3 -m venv .venv && source .venv/bin/activate   # Windows: .venv\Scripts\activate
pip install -r requirements.txt
pip install -e .
python scripts/run_game.py

That opens a random demo: the AI village manager picks valid actions at random (2D window). Add --human-3d for a passive 3D view (needs working OpenGL; on Linux/WSL you may need libgl1 / Mesa — if 3D fails, the script falls back to 2D).

---

Play against MAPPO

You control blue; red uses your checkpoint. Human play is 2D only (MAPPO training matches this).

python scripts/run_game.py \
  --mappo-opponent checkpoints/bots_mappo/mappo_bot_final.zip \
  --seed 0 --max-steps 500

Use a checkpoint trained with the same map size and game.max_bots_for_role_change as in your config.

---

Train MAPPO

python scripts/run_training.py

Quick test (small run):

python scripts/run_training.py \
  training.total_timesteps=2000 training.n_envs=1 training.selfplay_iterations=2

Settings live in configs/training/train_mappo_bots.yaml (merged via configs/default.yaml). Override anything under training.* on the command line.

TensorBoard: logs go under logs/mappo_bots/ when enabled. Then:

tensorboard --logdir logs/

Plots: python scripts/plot_tensorboard_scalars.py --log-root logs writes logs/plots/mappo_bots_scalars.png by default.

---

python scripts/evaluate.py

Runs many episodes with random village actions and prints rough stats.