Two years ago, Tesla CEO Elon Musk stood in front of a crowd at a specially-convened event at Tesla’s Hawthorne design studio and unveiled two brand-new features for Tesla electric cars: dual-motor all-wheel drive capability for Tesla Model S (and Tesla Model X); and a new set of hardware designed to make it possible for Tesla electric cars to drive themselves under certain situations.
The former gave Tesla’s electric cars more efficiency and better performance. The latter made it possible for Tesla to roll out its Autopilot semi-autonomous driver assistance technology. But while Tesla’s Autopilot hardware and software make the Tesla Model S and Tesla Model X one of the first cars to be capable of driving themselves with less human input than a car with no advanced vehicle hardware, Tesla’s first-generation Autopilot system isn’t capable of full autonomy — regardless of what some owners may or may not believe.
That is until just under an hour ago when Tesla announced that all Tesla electric cars being made from today onwards — including the upcoming 2018 Tesla Model 3 — will ship from the factory with second-generation Autopilot hardware capable of one day making full Level 5 autonomous driving possible.
As Tesla disclosed in an official blog post timed to coincide with the announcement, the new Autopilot V 2.0 hardware has been developed to work with Tesla’s own in-house autonomous vehicle software, and features eight surround-view cameras designed to give 360-degree visibility around the car at up to a quarter of a kilometer in all directions. These are combined with twelve updated ultrasonic sensors, allow the car to detect hard and soft objects at distances nearly twice that of the original Autopilot v 1.0 hardware, while a forward-facing radar compliments visual and ultrasonic sensors to help Tesla cars see through inclement weather.
Talking in a press call following the announcement, Tesla CEO Elon Musk said that he believes Tesla’s V 2.0 autopilot system to be “two times the reliability of a human driver,” adding that Tesla’s goal is to “become ten times as safe” as a human driver. That reliability, as with the rest of the autopilot system, will improve over time.
At the heart of Tesla’s new autopilot system is a brand-new computer system based around NVida’s Titan Graphical Processing Unit, a high-end massively-parallel chipset that’s usually found in high-end desktop and laptop computers. But while the Titan GPU is known for its high frame-rate in high-end gaming computers, it’s also ideal for just the kind of parallel processing that autonomous vehicle systems need in order to function safely.
We should note too that like the rest of Tesla’s Autopilot system, the computer system that will run Tesla’s proprietary autopilot software will be isolated from the rest of the car. The infotainment system computer — the one which operates the massive 17-inch touch-screen display found in the Model S and Model X — will be separate and distinct from this new high-end computer. As Musk joked in the press Q&A following the announcement “It would be bad if you could cause the autopilot to crash just by going to the wrong website!”
For those who are curious, Level 5 autonomy is defined as being one in which an autonomous vehicle can handle all of the dynamic driving tasks under all possible road conditions that a human driver could manage, without needing human intervention. In other words, a car which can be truly ‘hands off’. Currently, Tesla’s V 1.0 Autopilot hardware exists somewhere between Level 2 and Level 3 autonomy: the car can manage some of the dynamic driving duties (such as acceleration, braking and steering) but cannot deal with all road conditions or weather conditions and must rely on a human driver as a fallback measure.
To achieve full autonomy Tesla’s V 2.0 Autopilot hardware — termed “enhanced Autopilot” by Musk — features full redundancy in its sensor system, as well as all-round sensor heaters designed to ensure that snow or rain does not obscure the car’s view of the road ahead. While not specifically stated, the Autopilot computer system includes redundancy to ensure that it can always operate, since without it the car could not claim Level 5 autonomy.
While all Model S and Model X cars — and Model 3 — made from this point onwards will include all of the necessary hardware to make fully-autonomous Level 5 autopilot possible in the future, Tesla will continue the same policy when it comes to activating autopilot.
First, autopilot V 2.0-equipped cars will have the same kind of functionality as their V 1.0 counterparts, but as time progresses, Tesla will continue to roll out new features and capabilities, ultimately resulting in fully-autonomous capabilities (as soon as legislative and regulatory approvals have been met). Like V1.0 autopilot features, customers will be able to opt for a standard ‘free’ set of advanced safety features (where the car will act to protect itself in the event of an impending accident) or pay a premium for enhanced Autopilot activation to gain access to full autonomy.
Tesla says we’ll see the first demonstration of full autonomy from a V 2.0 equipped car next year, with Musk promising that a future demonstration of Tesla’s autonomous capabilities will include a coast-to-coast trip from Los Angeles to New York by the end of next year with zero human intervention behind the wheel or at the charging station. This means Tesla is not only working on an autonomous driving system but an autonomous charging system too.
As for indemnity against collisions involving Autopilot hardware? Thus far, Tesla’s Autopilot has been very reliable and Tesla says it sees no need to indemnify its customers against any accidents involving autopilot, choosing to leave that in the hands of the customer’s insurance company.
As Musk put it, Tesla’s autopilot system has to date, been very reliable, far more reliable than a human driver. But while there are some questions regarding the safety of such systems — and the legislative process involved in approving autonomous car technology, Musk was very clear about how he viewed any negative coverage of Tesla’s Autopilot system.
“When [the media] writes an article questioning autonomous car technology, [they’re] killing people,” he said.
In other words, Autonomous vehicle technology is something Musk sees as inevitable — and anything that stands in the way of autonomous cars should be questioned. We — and the rest of the automotive press — should consider ourselves duly warned.
You can also support us directly as a monthly supporting member by visiting Patreon.com.