Tesla's Full Self-Driving product is not what most people call 'full' self-driving

Tesla's use of the term doesn't match what the industry generally designates as autonomous vehicles that don't require human intervention.
Written by Tiernan Ray, Senior Contributing Writer

Tesla has started offering a subscription add-on for its automobiles for so-called self-driving features, though the language is a little confusing. 

Tesla just announced that it would sell a subscription for an enhanced version of what it calls "autopilot," called "full self-driving," for $199 a month, or $99 for those who are upgrading from the basic autopilot package. Full self-driving has already been available to customers for an up-front add-on of $10,000. The subscription offering simply means the plan can now be more widely accessible to drivers. 

The term is confusing, as the car using Tesla's software is not in complete control, suggesting that the 'full' in "full self-driving" is a misnomer. The term clashes with the language officially used in the language most people use to talk about autonomous driving, "advanced driver-assistance system," or ADAS.

As Tesla describes the service, full self-driving is when the car can "assist" in changing lanes; can help you to be a parallel or perpendicular park; will move from a "tight space" with just the click of the key fob or a tap of the app, or will automatically navigate a parking garage to you with a tap of the app.

There are a few additional features in beta, such as navigating on a highway or automatically slowing your car in response to traffic signals. One more feature, where the car automatically navigates city streets, is expected to be offered at some point in future.

This is confusing because none of this is that the human driver is still mostly in control. As the language on Tesla's site explains:

"The currently enabled Autopilot and Full Self-Driving features require active driver supervision and do not make the vehicle autonomous. Full autonomy will depend on achieving reliability far in excess of human drivers, as demonstrated by billions of miles of experience and regulatory approval, which may take longer in some jurisdictions. As Tesla's Autopilot and Full Self-Driving capabilities evolve, your car will be continuously upgraded through over-the-air software updates."

The ADAS designations used in industry, while being merely convention, tend to save terms of "full" for levels of steering a vehicle where a person is not doing anything and where a person may not be present at all in the vehicle.

For example, the Society of Automotive Engineers has defined the SAE J3016 standard for describing automated driving. There are actually six levels in SAE's definition, "six levels of driving automation, ranging from no driving automation (level 0) to full driving automation." 

The question of "full driving" automation is more extensively discussed in a 41-page document by the SAE, "Surface vehicle recommended practice." That document describes numerous details, frequently using the words "full" or "fully." 

For example, "full driving automation" is when "The user does not need to supervise a Level 5 ADS." Hence, a system such as Tesla's, where a person has to pay attention, is not the same use of the qualifier "full" as the industry usage, based on how the SAE uses the term.

In fact, Tesla's use of the term doesn't even really match level 4 of SAE's terms, "high driving automation." In level 3, under "full trip feature," the SAE specifies the ability for an automobile to be dispatched for a trip by a person with no one in the car:

EXAMPLE 1: A Level 4 ADS-DV is dispatched in driverless operation for purposes of providing ride-hailing services to customers located within its geofenced area of operation

Does it matter? To the extent there is already confusion among the public about what is autonomy in automobiles or other automated systems, misusing the term "full" is potentially an ethical issue in AI and machine learning.

One wall street analyst thinks Tesla may really clean up on the subscription offer. Pierre Ferragu of New Street Research estimates that there are 1.6 million Tesla's out there that are eligible, based on cars shipped since 2016 when the hardware to run self-driving became standard on Teslas. 

Ferrague offered a note to clients on Monday stating that, initially, there could be a negative financial impact for Tesla, as people switch from a paid add-on to a subscription plan. 

"On a quarter-by-quarter basis, the net effect will be negative. In the first quarter, about 30 subscribers would be needed to compensate for one up-front payer lost, from a cash perspective."

However, "We expect a growing contribution to gross margins," writes Ferragu, assuming 40-80% "penetration" of Tesla customers at some unspecified point in future. That might boost Tesla's gross profit by six to eleven points, sometimes around 2025, he writes.

Editorial standards