How Valeo’s validation team rebuilt its approach to ADAS testing and why a platform-first architecture changed everything.

Picture the moment your car gently corrects your lane drift at highway speed. Or the quiet alert that flashes when someone enters your blind spot. Behind every one of those interventions is an advanced driver assistance system (ADAS): a network of radar, cameras, ultrasonic sensors, and domain controllers that have been validated and reverified, and tested again under conditions as close to real-world as engineering can produce.

Valeo has been building and validating those systems for three decades. In that time, the company has shipped more than 1.5 billion ADAS sensors, and the pace is accelerating: Valeo expects to produce another 1.5 billion just in the next five years. By 2030, more than 90 percent of vehicles will include these systems. The scale is staggering, and so is the validation burden that comes with it.

“Each new part adds more to test. The system gets smarter, but validation gets harder,” says Vit Neruda, System Validation Manager for Valeo’s DETECT team. “As vehicles move to higher autonomy, ADAS content can grow up to 10X.”

Valeo’s DETECT team did not arrive at its current approach by design alone. It arrived by necessity—by running a well-functioning methodology all the way to its limit and recognizing the moment it could no longer scale.

Early validation methods worked. Data was logged in the vehicle and analyzed in the lab. Hardware-in-the-loop testing brought the real ECU into a simulated environment. Processes were documented. Coverage was solid. But as the ADAS system expanded, adding more sensors, more hardware, more signal paths, and more interactions to validate, the effort required grew proportionally—not just in engineering hours, but in the maintenance of bespoke test software developed alongside production software. Eventually, the team hit a wall that more resources alone could not solve.

“At some point, you cannot keep adding custom systems for every new requirement,” says Neruda.

The turning point came when the DETECT team asked a different question: not “how do we build better test benches?” but “how do we increase productivity without adding more overhead?”

The answer was architectural. Valeo standardized on NI PXI hardware as the foundation for direct injection HIL, a validation approach that uses the actual ECU running production firmware. That shift eliminated the need to build and maintain separate test software alongside production code, a significant source of overhead and risk.

With the same hardware base, the team can extend into simulation with different providers, offering flexibility to meet the varying requirements of different OEM customers. The platform became a common language across programs rather than a collection of program-specific one-offs.

The measurable impact showed up in two ways. First, platform reuse: the same PXI systems that go into the vehicle also operate in the lab, reducing variation and enabling setup reuse across programs. Second, performance: direct injection HIL using production firmware meant that validation results were directly reflective of how the production system would behave, delivered earlier, with more confidence, and without the lag of reconciling test and production software versions.

“Time to first test improved. Coverage improved. Engineering effort shifted back to validating the system,” says Neruda. “We did not scale by adding more custom work. We scaled by building on a common platform.”

As ADAS systems grow more sophisticated, artificial intelligence is becoming a larger part of the picture, particularly for perception and decision-making functions. Valeo’s approach to AI in validation is grounded in the same philosophy that shaped its platform shift: rigor first.

“These are safety-critical systems,” says Neruda. “We focus on validating AI with confidence. And we meet strict requirements for safety, security, and data privacy.”

That perspective, advancing capability while holding the line on safety, is the defining tension of automotive validation today. For Valeo, decades of experience building and validating ECUs form the foundation for bringing AI into production in a controlled and reliable way. The platform that enables scale also enables the discipline that safety demands.

What productivity looks like at this scale is not just faster. It is a validation approach that grows with the system without growing the burden.