Monday, May 22, 2024
by
Published
Views:
On April 20, Pacific AI Auto Analyst clarified that Tesla’s Full Self-Driving (FSD) system operates strictly as a Level 2 advanced driver assistance system (ADAS) in China under GB/T 40429-2021 — requiring continuous driver supervision. This regulatory framing now directly shapes compliance expectations for Chinese ADAS and sensor exporters targeting EU (UN-R157), South Korea (KC), and other regulated markets — especially regarding human-machine co-driving documentation.
On April 20, Pacific AI Auto Analyst stated that Tesla’s FSD system deployed in China is officially classified and registered as an L2 automated driving system per China’s national standard GB/T 40429-2021. All FSD functions require active, real-time driver monitoring and intervention readiness. Separately, the analyst noted that overseas regulators and customers — including those under UN-R157 (EU) and KC certification (South Korea) — increasingly require Chinese ADAS component and algorithm suppliers to submit a ‘China-version FSD human-machine co-driving logic manual’ as a mandatory attachment for market access review.
These companies face heightened technical documentation requirements when applying for foreign certifications. The demand for a localized ‘human-machine co-driving logic manual’ implies that hardware must be validated not only for performance but also for integration with region-specific supervision protocols — adding design and verification overhead beyond baseline functional safety testing.
Algorithm vendors supplying fusion perception or decision modules are now expected to formalize and document their driver engagement logic — including alert timing, takeover handover triggers, and fallback behavior — in alignment with China’s L2 operational definition. This affects both source code architecture and audit-ready documentation deliverables.
Integrators exporting full-stack ADAS solutions must ensure end-to-end traceability between sensor inputs, algorithm outputs, and driver interaction prompts. The ‘co-driving logic manual’ requirement signals that certification bodies no longer treat software as a black box — instead demanding explicit mapping from system state to human supervision expectation.
While GB/T 40429-2021 is a domestic standard, its operational interpretation — particularly around ‘continuous driver supervision’ — is being referenced abroad as a de facto benchmark. Companies should track how Chinese certification authorities (e.g., CATARC) articulate enforcement thresholds, as these may inform foreign reviewers’ expectations.
The ‘China-version FSD human-machine co-driving logic manual’ is emerging as a recurring request in EU and Korean pre-audit checklists. Exporters should treat this as a structured technical deliverable — covering supervision conditions, driver response validation methods, and failure mode escalation paths — rather than a one-off explanatory note.
Analysis来看, the current requirement reflects buyer-driven due diligence rather than a formalized international regulation. It signals growing scrutiny of human supervision logic — but does not yet constitute a binding clause in UN-R157 or KC texts. Companies should treat it as a leading indicator of future compliance expectations, not an enforceable standard at present.
From industry perspective, embedding documentation requirements into early-stage development (e.g., defining supervision states during architecture design, logging takeover readiness metrics during testing) reduces late-stage rework. Teams should map each functional requirement in their ADAS stack to a corresponding supervision logic statement — enabling faster manual generation and audit response.
Observation来看, this development is less about Tesla’s FSD status and more about how China’s L2 implementation framework is becoming a reference point for global ADAS governance. It reflects a broader shift: overseas markets are no longer solely assessing technical capability, but also how responsibility boundaries between machine and human are formally defined, tested, and documented. Current significance lies in its role as an early signal — not a finalized rule — indicating that ‘how supervision works’ is now as material as ‘what the system does’. Continued attention is warranted because such documentation expectations tend to cascade from high-profile OEMs (like Tesla) to tiered suppliers, eventually shaping certification templates and audit checklists.
This is not yet a standardized requirement across jurisdictions — but it is already a consistent pattern in commercial engagements involving Chinese ADAS technology exports.

Conclusion
Ultimately, the classification of Tesla FSD as L2 in China is not merely a local compliance footnote — it has become a functional anchor for international ADAS export readiness. For Chinese suppliers, it underscores that technical performance alone is insufficient; demonstrable, auditable alignment with human supervision frameworks is now a prerequisite for market access in regulated regions. This situation is better understood as an evolving operational benchmark — one that reflects growing global emphasis on transparency in automation responsibility allocation, rather than a fixed regulatory outcome.
Source Attribution
Main source: Pacific AI Auto Analyst, April 20 interpretation.
Note: The status of the ‘China-version FSD human-machine co-driving logic manual’ as a formal certification requirement — versus a common commercial request — remains subject to ongoing observation and is not codified in UN-R157 or KC regulations as of the reported date.

The Archive Newsletter
Critical industrial intelligence delivered every Tuesday. Peer-reviewed summaries of the week's most impactful logistics and market shifts.