Empiric Earth Proposes an Autonomous Truck Safety Test

Empiric Earth has collected roughly 10 billion miles of real-world driving data, giving the road-data company a rare-event base for a proposed safety test for autonomous cars and trucks.

Formed this year through the merger of Nexar and Nauto, Empiric Earth is developing a framework called Apex. Luc Vincent, the company’s special advisor, described Apex as functioning like a driver’s license for self-driving cars and autonomous trucks: an evaluation of performance in rare, dangerous scenarios rather than average conditions.

“You can then have a scorecard that a regulator or any kind of agency could use to decide whether it is safe to deploy some self-driving fleet or system on their roads,” Vincent said.

There is no universally accepted safety test before an autonomous vehicle operates on public roads, and freight has no equivalent evaluation standard for the software replacing human drivers.

That gap matters as autonomous trucking expands and debates intensify over commercial driver’s license requirements and carrier safety scores. Carriers have no sector-wide benchmark for comparing autonomous systems, while shippers and freight forwarders have no common basis for checking competing safety claims.

Vincent said the industry cannot wait for a legal crisis to create the standard. Apex could give carriers, shippers and freight forwarders comparable evidence and give regulators a shared reference for deployment decisions, rather than leaving each company to argue from its own test results.

What does Apex measure?

Vincent said most autonomous developers rely too heavily on simulation. He described the closed-loop environment as an echo chamber that lets performance drift over time, leaving real-world rare-event data to expose gaps virtual testing cannot fill.

Empiric Earth mines its 10-billion-mile archive for events that average mileage rarely captures:

• Near misses and road debris.
• Flooded streets.
• Jaywalking pedestrians.
• Wildlife crossings.

Apex would produce separate scores for individual scenarios and an integrated overall rating. The framework would also evolve as new road users and behaviors appear; Vincent cited scooters as a recent example.

Why do rare events matter?

Vincent put the scale this way: U.S. highway fatalities occur only once every 100 million miles. Observing events that rare, he said, requires extensive driving rather than a short test program.

Real-world exposure is what Empiric Earth says simulation cannot reproduce at the needed frequency. Targeted validation of those events, rather than simulation volume alone, is the company’s proposed answer.

Geography makes the issue more acute. The examples span:

• Austin’s horizontal traffic signals.
• Boston’s notorious left-turn behavior.
• Halloween pedestrian surges.
• Philadelphia’s unique traffic lights.
• Wildlife in Wyoming.

A system trained elsewhere may never have encountered a local signal layout, turning pattern, seasonal crowd or animal crossing when it enters a new operating area.

What changes for freight operators?

The commercial consequence is a shared reference point. Carriers could compare autonomous suppliers on one scale, shippers could review those providers on the same basis, and freight forwarders could apply a consistent safety comparison across vendors. Apex remains a proposed framework, not an adopted regulatory test.

For carriers, the score would sit alongside commercial driver’s license and carrier safety-score discussions as autonomous trucking expands. Shippers could use the same evidence during provider diligence, while forwarders could compare autonomous vendors without relying on separate, incompatible test claims.

Why is the timing urgent?

Vincent said the autonomous-vehicle market is accelerating. Multiple robotaxi fleets are now visible on San Francisco streets, while AI approaches are becoming more standardized and cheaper to implement. He said that growth makes formal testing more urgent, not less.

Vincent framed the risk: “As a result, I think it’s possible that in the future there will be more and more robotaxi companies… and at some point, when you have dozens or more, how do you know that developers haven’t cut corners and built something unsafe?”

Empiric Earth believes its rare-event dataset places it among the few organizations capable of building a credible, third-party benchmark for the industry. The next milestone is adoption: whether regulators, carriers and fleet operators accept Apex as a common deployment benchmark before expanding autonomous fleets make a standard unavoidable.