What Autonomy Level Is Your Car?

Classifications about the different stages of autonomy are confusing.


Over the past couple of months, you’ve probably heard semiconductor industry executives dropping numbers about the levels of autonomy for vehicles. And despite Tesla’s highly touted autonomous capabilities, current models are just a Level 2. Or maybe it’s a Level 3.

If these numbered levels were meant to lessen the confusion, it’s not clear the plan is working. Until last September, there were two independent classification systems for autonomous driving. The U.S. National Highway Traffic Safety Administration (NHTSA) had a five-level classification. The SAE had a six-level classification.

The NHTSA adopted SAE International’s six-level classification last September in an effort to eliminate that confusion. Competing standards can be a disaster, as the semiconductor industry well knows. And it’s worth noting that in each of these classifications there was/is a Level 0, which is where the driver does everything, so many people refer to the six levels of automation as five levels.

But within the now-accepted standard, there is growing disagreement brewing behind the scenes about whether Level 3 should be in there at all. At SAE Level 3, an automated system “can both actually conduct some parts of the driving task and monitor the driving environment in some instances, but the human driver must be ready to take back control when the automated system requests.”

While this sounds like progress, it’s hard to discern exactly what fits into levels 2, 3 and 4. This may sound like hair-splitting to most engineers, but lawyers and insurance companies have a completely different view of these definitions. If a driver is at fault, the driver is legally liable. If a car is at fault, the carmaker is liable. And if the definition is fuzzy, the likelihood is that both will be sued so the courts can decide.

That doesn’t bode well for the tech industry, which has never played a significant part in a mass-market safety-critical industry. For the most part, that burden has fallen on systems vendors/OEMs, which issue recalls and settle directly with government agencies and lawyers when something is faulty. But as more autonomous technology is added into cars, the burden shifts in unexpected ways. Both the logic and the sensors are a combination of internally developed and externally sourced components. If they are involved in serious accidents, there will be in-depth investigations into what went wrong and why.

This will entail far more detail than what goes wrong in a smart phone. A faulty part will be traced back through the entire ecosystem. Even if the culprit is a counterfeit part, legally that is someone’s responsibility. And given that all of this technology is brand new, that it will be updated over the air, and that there are always unforeseen corner cases even in well-understood applications, the path to autonomy will keep lawyers busy for quite some time.

All of this comes full circle back to definitions, which are difficult enough to develop in a static market. But as of three years ago, most car companies assumed that self-driving vehicles wouldn’t make their way into the market until 2030 or 2035. Carmakers have announced plans for fully autonomous vehicles to be on the road by 2020 or 2021, leaving the entire industry scrambling to come up with a blueprint. The NHTSA’s six-level standard, though, is hardly the kind of blueprint that will be required. It looks more like a slide that was hand-drawn while driving blindfolded over a rough road at high speed. And while it is spurring sales of technology at the moment, without a much more detailed roadmap a lot of companies could end driving off the road and getting ticketed, sued—or much worse.

Related Stories
Self-Driving Cars Rattle Supply Chain
Shift from hundreds of MCUs to more centralized computing is creating chaos in the auto industry.
What Can Go Wrong In Automotive (Part 3)
Why power has become so important in car electronics; the challenges in making autonomous vehicles reliable enough; adding margin for safe modes of operation.
Why Auto Designs Take So Long
A design for safety methodology is essential to trim automotive design costs, but at this point it’s a work in progress.