Tesla’s Cars are Beautiful — But They Aren’t Actually Self-Driving
Americans want safe self-driving cars, and Congress needs to act to unleash them
One of the most exciting things about self-driving cars is their ability to make our roads safer. As the National Highway Traffic Safety Administration (NHTSA) likes to note, 94 percent of serious crashes are due to human error, and self-driving cars could eliminate those accidents.
So why are Tesla’s cars — sold with “fully self-driving” capabilities — under investigation for causing crashes?
The straight answer is that Tesla’s cars — while beautiful and well-made — aren’t actually fully self-driving.
Tesla is under investigation because the software guiding its cars has trouble spotting parked vehicles. But in its fifty-plus year history, NHTSA has never before investigated a car for failing to “see” stationary objects.
The underlying issue here is that in case after case, Tesla’s drivers take their eyes off the road because they believe they are in a self-driving car. They aren’t.
Autonomy Levels
So what does fully self-driving actually mean? NHTSA places vehicles on a five-level spectrum of autonomous capabilities. Level 0 includes vehicles that are entirely human-controlled. Levels 1 and 2 have driver assistance features. Level 3 cars can drive themselves under select conditions, but drivers must be attentive at all times. Level 4 vehicles can drive themselves but with the option for a human driver to take over, while Level 5 cars do all the driving all the time.
When consumers think about fully self-driving cars, they imagine vehicles where they can take their eyes off the road, talk with loved ones, or even meditate. By NHTSA’s standards, those are Level 4 or Level 5 vehicles.
There are companies out there that are serious about developing fully autonomous, self-driving cars, and have deliberately taken a different approach. There are a few important differentiators that distinguish them from Tesla.
Benefits of Fully Autonomous Vehicles
First, transparency. Every major developer of autonomous vehicles from Cruise to Kodiak have submitted Voluntary Safety-Self Assessments (VSSAs) to NHTSA reporting on the safety of their automated driving systems. Some, like Waymo, have even released detailed safety whitepapers.To date, more than two dozen companies have published VSSAs on their self-driving cars. Tesla has never completed the voluntary report.
NHTSA also requires developers and operators of autonomous driving systems to submit more detailed crash reports than developers of advanced driver assistance systems like those found in Teslas.
Second, most companies developing full self-driving vehicles don’t just rely on a few cameras to keep their cars safely on track. Zoox vehicles use a suite of sensors, including cameras, lidar (light detection and ranging) sensors, and radar. Waymo uses a similar set of sensors to provide its vehicles with a 360 view with redundancies. Nuro’s self-driving bots use cameras, lidar, radar, and ultrasonic sensors.
Since May of this year, Tesla has been selling cars with driver assistance features based solely on eight cameras.
Last, but by no means least, the biggest developers of AV technology have taken a methodical, predictive approach to development of their product. They don’t bring something to market until they’re confident it’s fully functional and safe. They’ve tested millions of traffic scenarios with highly trained backup drivers. They’ve conducted testing in geofenced areas and select cities and under specific weather conditions to ensure safety.
In contrast, Tesla has taken an iterative approach to development. Although its vehicles still face challenges sensing stationary objects, Tesla is selling its autopilot feature to anyone who can afford it. Rather than developing a finalized product, Tesla is testing self-driving features on the public and making changes along the way.
The Federal Government Must Act
American consumers want safe self-driving cars, and policymakers should be doing everything possible to make sure these products are reliable. That means allowing responsible AV companies to begin putting safe vehicles on the road as they are finalized.
Right now, federal regulations limit AV companies to building 2,500 self-driving cars a year that don’t have human-driver features like a steering wheel, mirrors, or pedals. Those rules handicap the development of true Level 5 autonomy. Congress should ease those constraints, and federal regulators should finalize a long-awaited “occupant protection rule” for AVs. Allowing companies that have developed safe, tested AVs to put their cars on the road could save hundreds of thousands of lives over the long term.
Meanwhile, until its own vehicles meet a reasonable definition of “fully self-driving,” Tesla should consider the impact its marketing is having on drivers’ expectations. An iterative development might have worked for Elon Musk when he was designing rockets, but self-driving cars carry human cargo, and the stakes are higher.
The Chamber of Progress (progresschamber.org) is a new center-left tech industry policy coalition promoting technology’s progressive future. We work to ensure that all Americans benefit from technological leaps, and that the tech industry operates responsibly and fairly.
Our work is supported by our corporate partners, but our partners do not sit on our board of directors and do not have a vote on or veto over our positions. We do not speak for individual partner companies and remain true to our stated principles even when our partners disagree.