Sample Curators
Track owner manuals, feature pages, configuration tables, OTA notes, and government operating rules. Keep evidence dates and coverage metrics current.
OPEN STEWARDSHIP
OpenODC has a working first version, but it should not remain a one-person project. The next step is to build a real-time, credible, shared, and open platform with contributors who understand standards, ADAS / ADS development, testing, safety, data engineering, and open-source operations.
WHY NOW
The value of ODC data is not a one-off sample. It comes from continuously tracking owner manuals, feature versions, operating rules, regulatory material, and vendor-confirmed records. That requires domain expertise, standards discipline, and a sustainable maintenance model.
Track owner manuals, feature pages, configuration tables, OTA notes, and government operating rules. Keep evidence dates and coverage metrics current.
Maintain mappings between GB/T 45312—2025 and ISO 34503, ASAM OpenODD, BSI PAS 1883, SAE J3016, and related standards.
Review feature boundaries, fallback semantics, L2 / L3 / L4 wording, and the difference between feature availability and ADS responsibility.
Help OEMs, Tier 1s, test labs, associations, and standards organizations submit public ODC fields and vendor-confirmed records.
Automate source ingestion, evidence extraction, diff review, version updates, link checks, and publishing workflows.
Improve English terminology, international standard mappings, English pages, and collaboration with non-Chinese communities.
Assign evidence-as-of dates, source status, and owners to each sample so records do not become stale after OTA or manual changes.
Map owner manuals, PDFs, configuration tables, and operating rules into the 144 ODC elements with evidence snippets, page references, thresholds, and gaps.
Design corporate-domain verification, contact roles, public-field approval, sanitization notes, and admin review so vendors do not need GitHub.
Clarify merge criteria, error-correction timelines, reviewer roles, data licenses, and conflict-of-interest disclosure.
OpenODC describes public boundaries and evidence gaps. It does not score vehicles or companies.
Every key element should trace back to a manual, official page, operating rule, regulatory document, or clearly marked inference.
L2 records describe feature availability and limitations, not ADS responsibility for the dynamic driving task.
The code is open and the data is reusable, but high-confidence records require source, terminology, and semantic review.