Line Tracing — rule-based driving

No AI yet: you are the intelligence. The robot follows a line using rules you tune — and you'll feel exactly where rules reach their limit.

Run it

1. PhysiCar panel → Tutorials → Racing → course pill Rule-Based.
2. Click ▶ Line Tracing — it runs tutorials/racing-rb/line_tracing.py in the terminal. The app's MYAPP tab shows the camera and the color mask — white pixels are what the robot currently counts as "line".
3. Watch a lap. If it wobbles or loses the line, tune the rules (next section).

Tune the rules (the gear)

• Line color range (HSV) — widen or shift until the mask shows the line cleanly and nothing else. Lighting changes this; on the real robot you'll retune.
• Speed / steering gain — faster is harder: steering must react more per pixel of error.
• Camera pan/tilt — looking further ahead smooths driving, but reacts later.

Every change applies live — no restart.

What you just learned

This is rule-based AI: transparent, fast to build, and brittle — a shadow or a sharper curve breaks it, and you add rule upon rule. That pain is exactly why the next course lets the machine learn the rules itself → Supervised Learning.

Theory → Learn: the AI landscape