Genuine live policy session
Watch the driver decide.
This is not a recording. Your browser is running a fresh Python simulator plus the frozen v1.0 QR-DQN, RLAIF residual, and controller.
Waking the live driver
Loading Python and the parity-checked ONNX policy…
Elapsed
00:00.0
What is happening
One real policy decision every 100 ms.
Your browser runs NeonHighwayEnv-v5, forms the exact
33-value observation, and asks the frozen layered policy for one of
nine joint steering/pedal actions. No driving state leaves the device.
- Policy
- V5 QR-DQN + V6 RLAIF + V7/V8 controller
- Mode
- Endless; automatically restarts after a crash
- Scope
- Custom simulator only—not real-road autonomy