NAIT Teaching AI Simulator

A classroom-focused local app for teaching how a single-layer neural policy drives using ray sensors.

Highlights

• Cleaner simulation view with reduced visual clutter.
• Attempt-history traces so students can watch training evolution over many failures.
• Automatic restart after failure (default: 3 seconds, configurable).
• Context menu + options modal (manual numeric entry for settings).
• Updated figure-eight style track that avoids center-wall self-collision in 2D.
• Start/finish lines span lane walls (track-width crossing).

Core model

steering = tanh(bias + Σ(sensor[i] * weight[i]))

Setup

python -m pip install -r requirements.txt

Run

python app.py

Controls

• Space: pause/resume
• R: reset current attempt
• N: single-step one frame
• T / Y: next / previous track
• G: queue training batch
• C: toggle continuous training
• O: open/close options modal
• Right-click: open context menu

Attribution

See CREDITS.md for asset and license notes.