Tesla Quietly Upgrades Autopilot Hardware in New Cars

The automaker's hardware might not have been ready for full self driving after all.

If you’ve bought a Tesla, chances are its autonomous driving features helped to influence your decision on purchasing. As Tesla matured into the company it is today, its Autopilot division has continued to upgrade its hardware to support more self-driving functionality. According to sources, Tesla has taken a step up on its newest autopilot hardware, AP2, and has improved the hardware to address concerns over its ability to provide adequate processing power for future Level 5 self driving functionality.

One of the most confusing parts for a new prospective owner of a Tesla is understanding the different versions of autopilot available and which one they should purchased. Granted, from 2014 until 2016 Tesla only equipped vehicles with AP1, which utilized the MobilEye EyeQ3 (a recent acquisition by Intel)—so the decision was fairly easy. As the needs for more processing power became apparent, Tesla made the decision to discontinue AP1. In October 2016, any vehicle purchased going forward was upgraded to AP2, Tesla’s then latest-and-greatest hardware suite which would last to the implementation of full self driving. Now that somebody could snag a Model S used for a few thousand dollars less than a new one, the choice becomes a bit more difficult to choose.

If you jump in a car with AP1, you can expect to be able to drive down a highway and stay between the lines fairly reliably. Hands will need to stay on the wheel and the driver to remain attentive to the road. With AP2, once regulations are passed to govern Level 5 autonomy, a driver will become more of a passenger. This is where the additional hardware and processing power will come into play—more decisions require a more powerful computer. Currently, AP2 is regarded as behind AP1 in terms of functionality, though by the end of the year it is expected to surpass the now-defunct AP1.

But as Tesla began to develop AP2, they must have decided that there was a possibility their new hardware didn’t have enough power to handle the extra cameras and processing required. Thus, Tesla released a new version of their hardware, internally references as version 2.5, though internal sources feel that the changes don’t warrant that much significance. An additional Nvidia Pascal GPU was added for processing, as well as redundant camera wiring. The decision was likely made while under the leadership of Tesla’s former autopilot program, Chris Lattner, who once said that he wanted cars to be more like appliances. Lattner left in June after only six months in his position.

“However, we still expect to achieve full self-driving capability with safety more than twice as good as the average human driver without making any hardware changes to HW 2.0. If this does not turn out to be the case, which we think is highly unlikely, we will upgrade customers to the 2.5 computer at no cost,” a Tesla spokesperson said, according to The Verge.

Any new Tesla ordered today will ship with HW (“Hardware”) 2.5, though still designated as AP2. When full self driving is made available to Tesla vehicles, should an upgrade be necessary, it will be provided to owners at no cost. Tesla feels that this is very unlikely, but luckily is only talking about a small number of vehicles manufacturers from October 2016 until August 2017, less than a year of affected vehicles. If all goes well, no actions should be required from current AP2 owners.

Tesla still maintains that it will drive a car across the country without touching the steering wheel, though earlier this year announced it would delay the tests. It is possible that this new hardware development may have some correlation, though the reason for the delay was not specified. For some, this is seen as a broken promise from CEO Elon Musk. For others, it is a welcomed upgrade, so long as Tesla mitigates its notorious service center delays.