FAQ

Quick answers on product, data and deployment

Quick navigation:

NOA · Compliance · Testing & Validation · Competitors & Versions · Experience & Marketing · Insurance · Product & Download · Methodology & Scope · Glossary

NOA

Is NOA any good? How to evaluate?

  • Conclusion: trust long‑window multi‑city data, not one‑offs
  • How: ≥1,000 km/model, ≥3 cities, day/night/peak under one standard
  • Delivery: method notes, sample disclosure, replayable evidence

How to compare urban/highway coverage?

  • Conclusion: align samples first
  • How: slice by city/road/weather; unify windows and mileage
  • View: trends/heatmaps with sample sizes and windows

How many overrides are safe?

  • Conclusion: use hazardous km/override, not raw counts
  • How: scene thresholds (congestion/ramps), filter prompt‑only
  • Reference: disclose distributions and percentiles

How to count hazardous overrides?

  • Conclusion: automated + human; fully replayable
  • How: tag hard brake/steer/forced; align videos/trajectory
  • Review: human check edges; frame‑level replay/export

Compliance

How to test 2025 L2 compliance?

  • Conclusion: checklist, item by item
  • How: signal‑level DMS/ADAS + behavior‑level video
  • Output: per‑item pass rates + non‑compliant clips

How to validate DMS effectiveness?

  • Align eye/fatigue features with scene video
  • Quantify TPR/FPR by cohort/time/environment
  • Provide thresholds and tuning guidance

Testing & Validation

How to validate ADAS quickly?

  • Conclusion: templated, auto stats, closed loop
  • How: standard templates + detectors; auto pass/fail + causes
  • Landing: integrate ticketing (perception/planning/control)

Long‑mileage cost too high?

  • Conclusion: crowdsourcing + automation lowers cost
  • How: signed drivers; plan by sample×city×time
  • Scope: km‑scale samples for stable estimates

Is public‑road crowdsourcing compliant?

  • Conclusion: yes, with controls
  • How: signed drivers, RBAC, full audit trail
  • Before disclosure: anonymize/minimize

How to review corner cases?

  • Conclusion: find common failures; make a replay set
  • How: scene library + video slicing alignment
  • Focus: miss/double‑detect, rule conflicts

Can we replay issues?

  • Conclusion: fully replayable
  • How: sync interior/exterior video, sensors, event timeline
  • Landing: frame‑level RCA and exportable packages

Competitors & Versions

Compare competitor NOA directly?

  • Conclusion: yes; highlights and lowlights
  • How: unify by brand/model/version; radar/compare + clips
  • Scope: disclose sample sizes and windows

How to accept vendor stacks?

  • Conclusion: checklist acceptance with grading
  • How: multi‑vehicle sync capture; KPI checklist rating
  • Output: remediation list and re‑test plan

How to compare versions?

  • Conclusion: align samples, then compare
  • How: per model/version sample alignment; compare NOA/ADAS KPIs
  • Output: regression checklist and priorities

Experience & Marketing

How to extract launch selling points?

  • Conclusion: auto‑generate a selling‑point kit
  • How: representative KPI boards + visuals; one‑click exports
  • Scope: attach methods and sample disclosure

Use results in marketing?

  • Conclusion: yes, directly
  • How: one‑click chart/video exports for long/short content
  • Scope: attach sample/time/region; keep wording consistent

Why do reviews differ from KPIs?

  • Conclusion: comfort drives perception gaps
  • How: explain via accel curves and following variance
  • Landing: parameter/strategy tuning guidance

Insurance

How does data support UBI pricing?

  • Conclusion: risk features go straight into models
  • How: hazardous km/override; hard brake/turn features
  • Landing: segment by scene to increase discrimination

How do insurers identify low‑risk drivers?

  • Conclusion: use personas to find low‑risk groups
  • How: build scene/behavior‑based risk personas
  • Landing: incentives (points/cashback) reduce loss ratios

Product & Download

How to download and install the CHEK app?

  • On the Products page tap Android or scan the QR
  • iOS/IVI will be announced upon release

How is my data privacy protected?

  • Minimization, anonymization/minimization before disclosure
  • RBAC and full audit trail; itemized compliance acceptance

How to book a platform demo?

  • Use ‘Book a demo’ on the Products page
  • Or call us via the footer phone; we will arrange a walkthrough

How to become a partner / join the Open Alliance?

  • Click ‘Join now’ in the Open Alliance section to submit
  • Explore Open APIs links for data/integration options

How to contact support?

  • Phone: (86) 18410870301 (see footer)
  • Or leave a note via demo/partner forms and we will reach out

Methodology & Scope

  • Window: rolling 90 days by default; major releases split before/after.
  • Representativeness: slice by time (peak/off/night) and weather; cross‑city quotas.
  • Hazardous override: safety‑relevant overrides with hard brake/steer signature; prompts excluded.
  • De‑noising: remove device anomalies/missing footage; keep audit logs.
  • Auditability: every conclusion links to time window, route and clip IDs; third‑party witness supported.

Glossary

  • MPI (Miles per Intervention): mileage per override; higher is better.
  • Hazardous override mileage: average km per safety‑relevant override (km/override).
  • Coverage/Availability: share of time/mileage where NOA is operable under given conditions.
  • Success rate: proportion of scene‑defined goals (e.g., U‑turn, merge) successfully completed.
  • Comfort metrics: longitudinal/lateral acceleration distribution, following speed variance.