Updated and source-checked on February 27, 2026
Autonomous Vehicle Technology And Terminology (2026)
This glossary focuses on terms that materially change safety interpretation, regulation, and deployment analysis. Every definition is linked to a current primary source.
Direct Answers
What is the most common AV analytics mistake?
Treating all self-driving claims as equivalent. NHTSA separates ADS and ADAS event reporting, and its SGO page says incident data are not exposure-normalized.
What metric adds denominator context?
Exposure metrics such as miles and trips. Waymo publishes rider-only mileage and crash severity deltas, which helps contextualize event counts.
Terminology Table
| Term | Meaning | Why it matters |
|---|---|---|
| ADS (Automated Driving System) | In NHTSA crash reporting, ADS refers to SAE Level 3 through Level 5 automation modes that can perform the full dynamic driving task under defined conditions. | Readers should separate ADS claims from consumer driver-assistance claims, because reporting obligations and operational assumptions differ. |
| ADAS (Advanced Driver Assistance Systems) | NHTSA’s SGO framework separately tracks qualifying ADAS events (including systems marketed as Level 2), where the human driver remains responsible for supervision. | Conflating ADAS with ADS overstates real-world autonomy and can mislead safety comparisons. |
| ODD (Operational Design Domain) | An ODD defines where and when an automated system is designed to operate, such as specific roads, weather conditions, speed ranges, and geofenced zones. | Most 2025-2026 AV safety results are valid only inside a declared ODD and should not be generalized to all roads or weather. |
| FMVSS modernization tracks (102, 103/104, 108) | NHTSA announced in 2025 that it is pursuing AV-focused rulemaking updates for controls, windshield transparency and cleaning systems, and crash-avoidance standards. | This is a practical signal that AV policy moved beyond pilot language toward standards-level implementation work. |
| Event-based reporting versus exposure-normalized risk | NHTSA says SGO data are event-based and not normalized by miles traveled, making direct safety-rate comparisons incomplete without separate exposure data. | Good AV analysis combines incident reports with denominator metrics such as miles, trips, or insured years. |
2025-2026 Context To Keep In Mind
- NHTSA announced AV-focused FMVSS modernization tracks in September 2025 and released a final standards-research volume in December 2025.
- California reported more than 9 million AV test miles in its latest annual release, while NHTSA maintains national SGO incident updates on a monthly cadence.
- Waymo reports 127 million rider-only miles through September 2025 with lower injury-causing crash rates than its matched human benchmark in the same operating areas.
- NHTSA AV framework plan (opens in a new tab)
- NHTSA modernization update (Dec 2025) (opens in a new tab)
- California DMV AV annual mileage report (opens in a new tab)
- NHTSA SGO reporting dashboard (opens in a new tab)
- Waymo Safety Impact (opens in a new tab)
Continue with Laws and Regulations for policy specifics, or review the FAQ for concise rider-facing guidance.