Run Library

Runs recorded at testing sessions, replayed in the browser. A decision replay steps through every scan the car saw and explains the steering and speed it chose; a map replay shows the course it mapped and the raceline it planned.

Decision replay Car 2, gap follow, October 19, 2026 2026-10-19 20:07 35.4 s driven 78.2 m 3.75 m/s top 7 stops Open the replay Map and raceline Car 2, mapped course and raceline, July 27, 2026 2026-07-27 20:01 26.3 m lap 2.47 m/s top Open the replay
How a run gets here

Runs recorded at testing sessions are archived in the car workspace's data/ folder on GitHub. Two exporters turn one into a page here.

A gap_follow run recorded with race_diagnostics, after tools/run_analysis/analyze_run.py has cached its bag, becomes a decision replay. It needs numpy, PyYAML and a checkout of the car workspace:

python3 scripts/export-decision-run.py <workspace>/data/<id>/racerbot_runs/<run-dir> \
    --workspace <workspace> --yaml <gap_follow.yaml the car ran> --car 2 --site .

A mapping run with a saved map and raceline needs only Python 3:

python3 scripts/export-run.py <workspace>/data/<id>/racerbot_auto/<run-dir> --site . --title "..."

Each writes its data to static/runs/<run-id>/ and a page under content/resources/runs/. Commit both and push, and the replay is live.