Neuroevolution (rust + bevy + avian2d)

Neuroevolution Simulator (work in progress) writing using rust language. It shows the usage of genethic algorithm, NEAT (Neuroevolution of Augmented Topologies) in unsupervised machine learning methods.

Alife

Introduction
Evolve 2 is a simulation program written in rust language with Bevy Game Engine and Avian2D physics engine. The aim of the simulation is trying to model some kind of environment and multiple agents interacting themselves and changing the environment dynamically.


Environment
Environment is a grid structure reflecting terrain height and water level. Terrain height is created based on perlin noise. Water level is dynamically updated between single grid cells.


Plants
Semi-active agents without neural network, but with evolution based physical attributes. They can grow, dye, multiply, inherit attributes, and evolve.


Agents
The crucial elements of simulation. Agents have physical attributes and neural networks based on genetic algorithm. They can grow, dye, multiply, inherit attributes, and evolve.