🏠 Home

NEUROEVO

Watch AI rockets evolve to reach the green target. Each rocket has a tiny brain (neural network) that learns over generations — no programming, pure evolution.

Drag the green target to move it

Use Draw Wall button to add obstacles

Vision lines show what the best AI "sees"

The brain panel shows neural activity live

On mobile: use bottom control bar. On desktop: Shift+drag to draw walls.

NEUROEVO

Neural evolution live

GEN 1

Drag target  |  Shift+drag walls  |  Space Pause

AI Thinking — Live
WAITING First generation starting...
Stats
1
Gen
40
Alive
0
Reached
Best Fitness
0.00
Avg Fitness
0.00
0% of generation
Best Brain — Live
Positive
Negative
Active
Learning Curve
Scenario
Speed
Mutation Rate
5%
StableChaotic
How AI Learns

1. Brains — Each rocket has a neural network: 6 inputs (eyes), 8 hidden neurons (thinking), 2 outputs (thrust direction).

2. Evaluation — After each generation, fitness is scored by how close each rocket got to the target.

3. Selection — Top performers are picked as parents. Better fitness = higher chance.

4. Crossover — Child DNA is a mix of two parents' brain weights.

5. Mutation — Small random changes add new traits. Over many generations, navigation strategies emerge naturally.