How the AI works
A genetic algorithm evolves 200 walkers. Each individual is a fixed sequence of moves (its "genome"); there's no neural network here — just selection on sequences.
Genome and fitness
- Genome: a list of directional moves.
- Fitness: how close the walker gets to the exit (and how quickly).
Evolution loop
The closest walkers are selected, their move sequences are crossed over, and random mutations tweak a few steps — so each generation drifts toward solving paths.
What you see on screen
Pheromone-style trails glow on frequently used cells, revealing the promising routes the population is converging on.