The transparency website PlainSite recently published a pair of letters that Tesla wrote to the California Department of Motor Vehicles in late 2020. The letters cast doubt on Elon Musk’s upbeat timetable for developing completely driverless technology.
For years, Elon Musk predicted that completely driverless technology is just around the corner. At an April 2019 event, Musk predicted that the Teslas would be able to operate entirely without a driver – known in industry jargon as “tier 5” – by the end of 2020.
“There are three steps to driving alone,” Musk told Tesla investors at the event. “There are complete resources. Then, complete resources to the extent that we think the person in the car does not need to pay attention. And then, we are at a level of reliability where we also convince regulators that this is true.”
Tesla obviously missed Musk’s 2020 deadline. But you can be forgiven for thinking that Tesla is now running late on the strategy he described two years ago. In October, Tesla launched what it called “fully autonomous beta” software for a few dozen Tesla owners. A few days ago, Musk announced plans to expand the program to more customers.
Given that the product is called “fully autonomous driving”, this may seem like the first step in Musk’s three-step progression. After a few more months of testing, it may become reliable enough to operate without human supervision. This could allow Musk to stick to his latest optimistic schedule for autopilot: in a December 2020 interview, Musk said he was “extremely confident” that Tesla vehicles would reach level 5 by the end of 2021.
But a letter that Tesla sent to California regulators in the same month had a different tone. Despite the name “fully autonomous”, Tesla admitted that it does not consider the current beta software suitable for fully driverless operation. The company said it would not start testing “truly autonomous features” until some point not specified in the future.
“We do not expect significant improvements”
In two letters last November and December, California DMV officials asked Tesla for details on the FSD’s beta program. Tesla requires drivers who use the beta software to actively supervise it so that they can intervene quickly if necessary. The DMV wanted to know if Tesla planned to relax human supervision requirements once the software was made available to the general public.
In its first response, sent in November, Tesla emphasized that the beta software had limited functionality. Tesla told state regulators that the software “is unable to recognize or respond” to “static objects and road debris, emergency vehicles, construction zones, large uncontrolled intersections with multiple entry routes, occlusions, adverse weather, vehicles complicated or adversaries in the path direction and unmapped roads. “
In a December follow-up, Tesla added that “we expect the functionality to remain largely unchanged in a future full launch for the customer’s fleet.” Tesla added that “we do not expect significant improvements” that would “shift responsibility for the entire dynamic steering task to the system.” The system “will continue to be an SAE Level 2, an advanced driver assistance feature”.
SAE level 2 is industry jargon for driver assistance systems that perform functions such as lane maintenance and adaptive cruise control. By definition, level 2 systems require continuous human supervision. Fully driverless systems – as the Waymo taxi service is operating in the Phoenix area – are considered to be level 4 systems.
In his letter to California officials, Tesla added that “the development of true Tesla autonomous features will follow our iterative process (development, validation, early release, etc.) and any of these features will not be released to the general public until we have fully validated them. “
Critics took advantage of the disclosure. “Here it is, straight from Tesla,” tweeted the prominent Tesla skeptic Ed Niedermeyer. “‘Complete self-steering’ is not, and never will be, really self-directed.”
This may not be very fair for Tesla – the company apparently plans to develop more advanced software in the future. But at the very least, Tesla’s public communication about the complete standalone package could easily give customers a misperception about the software’s future features.
Total autonomy is always just around the corner
Since 2016, Tesla has given customers every reason to expect their “fully autonomous” software to be, say, fully autonomous.
The first promotional materials in the FSD package described a driver getting out of the vehicle and making him find a spot on his own. Tesla has been talking repeatedly about the FSD package that allows a Tesla vehicle to operate as an autonomous taxi – an application that requires the car to drive alone with no one behind the wheel. In 2016, Musk foreseen that, within two years, a Tesla owner in Los Angeles would be able to summon his New York City vehicle.
If Tesla is really going to achieve fully driverless operation in 2021, that won’t leave much time to develop, test and validate complex, security-critical software. Therefore, it would be natural for customers to assume that the software that Tesla called “Full Self Driving beta” is actually a beta version of Tesla’s long-awaited fully autonomous software. But in its communications with California officials, Tesla makes it clear that this is not true.
Of course, Elon Musk has a long history of announcing timelines that are too optimistic for his products. Unsurprisingly, Tesla has failed to meet an optimistic deadline set by its CEO.
But there is a deeper philosophical issue that can go beyond some overdue deadlines.
The long road to full autonomy
Waymo
The general strategy of Tesla’s autopilot is to start with a driver assistance system and gradually evolve it into a completely driverless system. Several other companies in the industry – led by Google’s Waymo – believe this is a mistake. They think the requirements of the two products are so different that it makes more sense to create a driverless taxi, space shuttle or delivery service from scratch.
In particular, companies like Waymo argue that it is very difficult to get regular customers to pay close attention to an almost driverless vehicle, but not entirely. If a car drives perfectly for 1,600 kilometers and then makes a big mistake, there is a significant risk that the human driver is not paying enough attention to avoid an accident. Waymo initially considered creating a driver assistance system similar to autopilot and a license for car manufacturers, but the company decided that this would be too risky.
Musk has always ignored this criticism. As we have seen, he believes that the improvements in Autopilot’s driver assistance features will transform it into a system capable of operating entirely without a driver.
But in its comments to the DMV, Tesla seems to endorse the opposite view: that adding “real standalone features” to Autopilot will require more than just incrementally improving the performance of its existing software. Tesla recognized that it needs more sophisticated systems to deal with “static objects, road debris, emergency vehicles, construction zones”.
And that makes it a little hard to believe in Musk’s boast that Tesla will reach level 5 of autonomy by the end of 2021. Notably, Google’s autonomous vehicle prototypes have been able to navigate in most road conditions – very similar to today’s Tesla FSD beta software – since approximately 2015. Still, the company needed another five years to refine the technology enough to allow for completely driverless operation.
And that was within a limited geographical area and with the help of powerful sensors to deal with. Tesla is trying to achieve the same feat for every street in the country – and using only cameras and radar.
Maybe Tesla will move faster than Waymo and it won’t take another five years to achieve completely driverless operation. But customers who are considering paying $ 10,000 for Tesla’s full standalone software package should certainly understand Musk’s upbeat schedule with a pinch of salt.