By ATN Editorial Team
Summary: Nexar has launched Nexar Apex, the first real-world credibility benchmark for autonomous vehicles, replacing simulation-based assumptions with standards grounded in billions of human-driven miles. Built on Nexar’s Real-World Data Engine — which captures over 100 million fresh miles monthly — Apex evaluates AV performance against human behaviour using Nexar’s BADAS model, trained on millions of collisions and near-misses. Nexar also introduced the AV City Readiness Index, which assesses how suitable different cities are for safe AV deployment based on real-world risk indicators and only in locations where Nexar maintains 99% road-coverage accuracy. Together, Apex and the Index offer a unified, data-driven framework for regulators, insurers, cities and AV developers to measure true readiness for autonomous driving on public roads.
Key engineering takeaway: Nexar Apex uses a massive real-world driving corpus and the BADAS human-behaviour model to directly compare AV system outputs to human responses, providing the industry’s first scientifically grounded, real-world benchmark for AV safety.
Why it matters: By anchoring AV evaluation to real human behaviour instead of simulation alone, Nexar gives regulators, insurers, and developers an objective way to measure readiness and determine where AVs can be safely and confidently deployed.
Nexar, a leader in AI-powered mobility solutions and one of the largest distributed vision networks on U.S. roads, today introduced Nexar Apex, the first real-world credibility test for autonomous vehicles. Built on Nexar’s Real-World Data Engine — which captures more than 100 million miles of fresh road data every month — Nexar Apex replaces abstract simulation-based assumptions with measurable, human-grounded performance standards for evaluating when an AV is truly ready for public roads.
For years, autonomy has been shaped by two contrasting sources of intelligence. Artificial Intelligence is built on simulated worlds and estimated probabilities. Physical Intelligence comes from observing how people actually drive in real conditions — the chaos, surprise, and nuance that no synthetic environment can fully replicate. Simulation helps train models, but it cannot define safety on its own. Only real-world data, measured continuously at massive scale, can anchor the industry in reality.
“Simulation is a training tool, not a proving ground,” said Zach Greenberger, CEO of Nexar. “To make autonomy operationally scalable, we need a benchmark rooted in Physical Intelligence — knowledge born from billions of real miles and actual human decisions. Nexar is now providing the industry with the definitive ground truth needed to move from theoretical safety to trusted deployment.”
The “Miles-to-Confidence” Gap
The AV industry has relied heavily on simulation environments to approximate safety. While useful, these struggle to reflect the full spectrum of ground-truth driving behaviors — the erratic lane-changes, unpredictable pedestrians, sudden construction zones, weather anomalies, and thousands of real-world edge cases that define physical roads. This gap is not academic; according to industry research, proving that an autonomous vehicle is safer than a human driver with strong statistical confidence requires hundreds of millions to billions of real-world miles. Without a shared, objective yardstick rooted in reality, insurers, regulators, and cities have had no practical way to evaluate readiness.
Nexar Apex changes that by anchoring AV safety to the truest source of physical intelligence available: Nexar’s 10-billion-mile corpus of real-world human driving.
The Shift to Physical Intelligence
At the core of Nexar Apex is the industry’s first open measurement for machine performance. Safety is relative — to judge an autonomous system, you must compare it to the human it aims to replace.
Accessible via Nexar’s API, the framework allows developers to query their stacks against Nexar’s BADAS model. Trained on millions of collision and near-miss events, this model quantifies human reaction times, anticipation patterns, as well as danger-avoidance behavior. By comparing AV outputs to this data set of ground-truth driving behaviors, Nexar Apex establishes a scientifically rigorous standard, giving legislators, insurers, developers and, most importantly, the public, an aligned way to understand the risks and rewards of an AV model.
A Data-Driven AV City Readiness Index
Alongside the new testing standard, Nexar introduced the AV City Readiness Index, a first-of-its-kind model that measures how ready a city is for safe and economically viable AV deployment.
A mile in Phoenix is not a mile in Boston; operational complexity varies dramatically by geography. Nexar normalizes these differences using objective indicators derived from its Real-World Data Engine, including collisions per million miles, harsh-braking density, construction zone volatility and other predictive signals that proactively identify risk before incidents occur.
Setting the Standard for Data Density
To ensure scientific rigor, Nexar only evaluates cities where its network maintains a Ground Truth Threshold of at least 99% road coverage and monthly data freshness. This ensures the Index reflects cities as they exist today — not in static maps or outdated datasets — giving operators, regulators, and insurers a clear view of where AV fleets can scale with confidence.
Turning Real-World Ground Truth Into Industry Standards
Nexar Apex and the AV City Readiness Index together form the first unified framework that brings objective clarity to the “miles-to-confidence” problem. Nexar invites AV developers, insurers, regulators, and city partners to access the standards, explore readiness scores, and learn how real-world data can accelerate trusted deployment.
For more autonomous vehicle technology news, click here.
