A 2-player competitive game where you don't fight directly — you design robot bodies, evolve their brains through neuroevolution, and watch them battle autonomously.
Build the body. Evolve the brain. Watch them fight. Adapt.
This project is a work in progress and not yet playable as a full game. Core systems like robot assembly, neural network training, and the genetic algorithm are functional, but major features (resource economy, battlefield combat loop, scanning/arms race, win condition) are still missing.
- Robot Assembly — modular block-based robot designer with armor, engines, weapons, sensors, radar, beacon, scanner, and gatherer modules
- Neural Networks — small feedforward nets with auto-derived inputs/outputs based on robot design, live network visualization during assembly
- Neuroevolution — genetic algorithm training in parallelized subprocesses, configurable fitness functions, hot-swappable training zones
- Physics — vectorized NumPy-based sensor calculations, collisions, and movement
- Resource system (income, spawning, recycling)
- Battlefield combat loop and win condition
- Scanning enemies for use as training dummies
- General polish and balancing
Python, Pygame, NumPy