Pins, wall slide, and first lap-gate harness

· Dirt Bowl Rally (offroad)

Progress

First serious push toward a one-lap gate: scripted Puppeteer run from the grid with timed W / A / D segments. By end of day the harness reports about 57% of track arc-length (tp ≈ 2312 / 4042) with laps=0.

Learnings

Centerline ≠ midpoint between barriers. Progress is arc-length on a BFS centerline through dirt; the drivable mask is a tube minus barrier pixels. Misalignment shows up as “on track” in telemetry but wrong line visually — filed as centerline drift work (t-f1f9).

Wall contact should bleed speed along the tangent, not kill forward motion when scraping parallel barriers. Lesson captured in t-lesson-wall-slide.

Feel target: Super Off Road reference lap @ 0:44 (t-design-super-offroad).

Failures

Gate still false — no laps→1 despite tp jumping mid-run (e.g. 668→2312 in one W segment). That jump is partly projection/crawl integration, not pure driving distance (t-513b).

Hairpin and south-wall pins blocked early loops; after fixes, the bottleneck became arc-length stability and lap counting.

Screenshots

Two frames from shots-verify-20260524-221944 — full sequence left in /tmp/offroad-play/.

Truck scraping a red and white striped barrier
Parallel wall scrape — why we added center resolve + tangent slide (bug-south-wall-pin, t-lesson-wall-slide).
Trucks on a narrow dirt corridor between striped barriers
Drivable corridor vs barrier paint — centerline/mask alignment is the long-term drift issue (t-f1f9).