Racing — one track, every AI method

The Racing courses are plain Python files in your workspace, ~/physicar_ws/tutorials/racing- — installed and kept up to date by the PhysiCar panel › Tutorials tab › Racing, which runs them for you from the terminal. You will climb the ladder of AI methods on the same track:

1. Line Tracing — rule-based: you write the rules (HSV or RGB color + steering), no training at all. Files: tutorials/racing-rb/line_tracing.py + tutorials/racing-rb/settings.json (+ myapp.html, the page the course shows in the MYAPP tab while it runs).
2. Supervised Learning — you label photos, a CNN learns to imitate you. Files: tutorials/racing-ml/train.py, tutorials/racing-ml/inference.py + tutorials/racing-ml/settings.json (+ myapp.html); labeled photos in tutorials/racing-ml/data/{left,straight,right}/, trained models in tutorials/racing-ml/models/<id>.onnx.

How a course page works

• The ▶ button (Line Tracing / Train / Inference) types the course command into the terminal — python3 ~/physicar_ws/tutorials/racing-rb/line_tracing.py, python3 ~/physicar_ws/tutorials/racing-ml/train.py or python3 ~/physicar_ws/tutorials/racing-ml/inference.py — and the output is the terminal. The live view is the app's MYAPP tab while a course runs.
• Stop is Ctrl+C in that terminal, or the panel's Stop button.
• The gear holds the course settings — speed, steering, camera angle, line color — stored in the course's settings.json (edit either one) and applied live to a running course.
• The scripts are in the explorer under tutorials/ (read-only). Read them when you're curious; copy a script into your own folder when you want to go custom.
• One course drives at a time — starting a run stops the other. Training can run alongside a drive.

Race day? Time your laps with the Racing Lap Timer and start fair with the Remote Traffic Light.