A joint framework from Lockton and Nexar

Autonomous vehicle safety, measured against real human driving

An independent benchmark from Lockton and Nexar that shows whether an AV performs at, above, or below a human driver in the same conditions, grounded in how people actually drive.

Benchmark your AV system →Talk to our team about AV risk

Every safety claim needs a baseline. The industry never agreed on one.

For decades, road safety has been measured against human performance. Fatalities per mile, incident rates by geography, near misses by weather and time of day.

When autonomous vehicles arrived, the obvious question followed: Is the machine safer than the person?

That question has been hard to answer credibly. Most AV safety data is measured against baselines that were never designed for the comparison, so results are difficult to trust and nearly impossible to compare across companies. The benchmark this framework provides is the missing piece: a consistent, independent reference for how AVs perform relative to real human drivers.

What the framework does

It answers one question, in context:

does this AV system perform at, above, or below a human driver in the same conditions?

The benchmark is built on observed human behavior rather than assumptions about how people are supposed to drive, and it compares systems within relevant operating environments rather than across generalized scenarios.

It is designed as an open, evolving reference, not a proprietary rating, and built to serve three audiences at once: insurers structuring coverage, regulators evaluating safety claims, and AV developers proving performance to both.

0:00
/
0:00
576p

How it works

The framework is powered by BADAS 2.0, Nexar’s collision anticipation model family, trained entirely on real-world driving with no synthetic data. Nexar’s network of more than 350,000 cameras, covering 94% of U.S. roads, provides the scale that makes the benchmark meaningful rather than anecdotal. It works in two parts.

Nexar Risk Index

An environmental risk assessment that evaluates an AV’s operating domain, accounting for geography, road conditions, and driving profile. Context is established before any comparison is made, so an AV in dense urban traffic is not measured against the same baseline as one running highway logistics.

Nexar Apex

A submission platform where AV developers test their models against curated real-world edge cases drawn from comparable environments. The output is direct: performance relative to a human baseline in defined conditions.

What it changes, depending on where you sit

If you build AVs

A credible, third-party way to demonstrate safety in a format insurers and regulators can actually use, without exposing proprietary data to competitors.

If you underwrite risk

A way to close the gap between AV deployment and measurable risk, with environmental context built in before any system is priced.

If you regulate or set policy

A reference grounded in observed behavior rather than manufacturer-reported metrics aligned with how road safety has always been measured.

Built on established practice

Nexar data has already been used by leading AV developers to support published safety cases and peer-reviewed research. This framework formalizes approaches already in use at the highest levels of industry scrutiny, with Lockton’s risk expertise and Nexar’s data infrastructure extending them to the broader market. It is not a finished system. As deployment expands, the dataset deepens and the comparisons sharpen.

See where your system stands

The benchmark is no longer theoretical. Submit your model to Nexar Apex and see exactly where it performs against the human baseline in the environments you operate in. Evaluating AV exposure across a portfolio or jurisdiction? Talk to our team!

Benchmark your AV system →Talk to our team