Self-Driving Car Simulator
Cars learn to drive by themselves using a neural network — a simple artificial brain inspired by how real neurons work.
How it works
- Sensors — Each car has yellow rays that detect obstacles and road edges. These are the brain's "eyes".
- Brain — The neural network (shown on the right) takes sensor data and decides: go forward, turn left, turn right, or reverse.
- Selection — The car that gets the furthest without crashing is the "best" (shown in full color). All others are transparent ghosts.
- Evolution — Save the best brain, restart, and all new cars get slightly mutated copies. Over generations, they learn to drive!
Training workflow
- Watch cars drive. The best one is highlighted automatically.
- When you're happy with one, click 💾 Save to remember its brain.
- Click 🔄 Restart — new cars spawn with mutations of the saved brain.
- Repeat! Each generation should get better.
- Click 🗑️ Discard to start completely fresh.
Neural network display
The black panel on the right shows the best car's brain in real-time:
- Bottom nodes — sensor inputs (how close obstacles are)
- Middle nodes — hidden layer (the "thinking" part)
- Top nodes — outputs: ↑ forward, ← left, → right, ↓ reverse
- Yellow lines — positive connections (encouraging)
- Blue lines — negative connections (inhibiting)
- Dashed circles — neuron bias (activation threshold)
Buttons
| 💾 | Save the best car's brain |
| 🗑️ | Delete saved brain, start fresh |
| 🔄 | Restart simulation (keeps saved brain) |
| ⚙️ | Open settings panel |
| ❓ | This help screen |