This project is a C++ predator–prey simulation where agents are controlled by neural networks trained through evolutionary methods. Predators and prey interact in a simulated environment, and behaviors emerge through selection and mutation over generations.
The project is inspired by the final project of a course I took at IMSA called Computational Science, where the entire class wrote deterministic or MLP-based policies for lynx and hare and competed in a “Hunger Games” style arena to see whose hare would survive the longest.
Work in progress
- Raycasting for agent "vision"
- Spatial partitioning for collisons/rays
- Neural Network for agent "brain"