Teslas now being produced with full self-driving hardware

Tesla self-driving hardware

Tesla’s latest revelation is out, it’s a big one, and this time everyone should be able to appreciate its significance. In less than two years, all Tesla models should be capable of full level-5 autonomy, driving themselves without human input. Tesla has leapfrogged far, far beyond every other automaker to introduce an innovation that could radically change the way we get around, and even the way cities and road networks are designed.

Here’s the gist of it, from the Tesla blog: “…as of today, all Tesla vehicles produced in our factory – including Model 3 – will have the hardware needed for full self-driving capability at a safety level substantially greater than that of a human driver.”

At this point, only the hardware part of the system is in place. The software that makes it all work will be enabled gradually via over-the-air updates. In fact, the new hardware suite will not offer the same Autopilot capabilities that the old one does at first, but it should reach parity after a few months of software improvements. The software will be operating in “shadow mode,” constantly gathering data to improve the system, until it is capable of full autonomy by 2018. Elon Musk promised a demo of a fully autonomous drive from Los Angeles to New York by the end of 2017.

Tesla described the new hardware suite: “Eight surround cameras provide 360 degree visibility around the car at up to 250 meters of range. Twelve updated ultrasonic sensors complement this vision, allowing for detection of both hard and soft objects at nearly twice the distance of the prior system. A forward-facing radar with enhanced processing provides additional data about the world on a redundant wavelength, capable of seeing through heavy rain, fog, dust and even the car ahead.”

tesla-self-driving-hardware tesla-self-driving-hardware tesla-self-driving-hardware

“To make sense of all of this data, a new onboard computer with more than 40 times the computing power of the previous generation runs the new Tesla-developed neural net for vision, sonar and radar processing software. Together, this system provides a view of the world that a driver alone cannot access, seeing in every direction simultaneously and on wavelengths that go far beyond the human senses.”

The Tesla Neural Net relies on the Nvidia Titan GPU, but doesn’t use any third-party hardware sensors, Musk said in a conference call following the announcement. The new system won’t be sold to other automakers, as it would be “very hard to turn into a kit.”

No, all the wonderful new goodies will not be free. The safety features enabled by the system will come standard on all models, but the Enhanced Autopilot package will be a $5,000 option, and Fully Self-Driving mode will cost $8,000.


Source: Tesla, TechCrunch, The Verge, Electrek

  • brian_gilbert

    Does not solve the human driver problem whereby a human driven car ccrashes into a perfectly innocent autonomous car which can ne stationary at the time. A completely driverless zone only requires a driverless vehicle to srick to its road lane and that is viable now.

    • Lance Pickup

      Might be viable, but requires a massive, sudden shift in all vehicles to driverless (and how viable is that today?), or the creation of driverless zones which would similarly require a massive investment in vehicles to service those zones, and possibly require drivers to get out of their driven vehicles at the edge of the zone and move into driveless vehicles. While “viable”, I think you would certainly meet with intense resistance to that approach. All this to solve a problem that is not zero probability, but is not rampant either. I applaud your enthusiasm for shifting to a completely driverless transportation system, but I think we need to be realistic and recognize that such a step change is not practical and we will have to have driver and driveless co-exist for quite some time. Don’t let perfect be the enemy of good here.

      • brian_gilbert

        – Investment in vehicles
        The increased capacity of the existing roads renders unnecessary budgetted expansion for the next decade or so. The implementation would be in progressively increasing size stages to reveal changes needed to the plan. So plenty of money available immediately followed by the increasing savings at each stage. At most 10% the number of vehicles required for the zone and we already renew more than that every year.
        -Drivers would meed to switch vehicles at the edge of the zone. London has managed with a congestion zone for some years with no great protest. I would choose a location like the Isle of Wight for the first location so that switching coincided with the sea crossing.
        – To solve a problem that is not rampant:
        It is rampant at present in London. There are many collisions and many of these create a traffic jam. The Managing Director of traffic for London just said that on a tv program viewing congestion in many places. The average speed is less than 10mph. Cyclists can beat motorised vehicles frequently. They do so at the cost of mamy of their lives and the cost of the memtal wellbeing of the driver that kills them.
        – Have drivers and driverless coexist:
        But they can’t. Reports of crashes between drivers and driverless seem always the fault of the driver and yet are considered a reason for not going driverless. Drivers can never be expected to be completely safe. On that TV program I mentioned one happily admitted that he broke the law when he considered there was little risk. We know that many people cannot stop themselves using phones surreptitiously. We know that when people are in a hurry they take chances. People frequently drive while over the alcohol limit. A small risk on each case but 1.2 million are killed on the worlds roads every year.
        I have been studying the idea for over a year and committed my findings to a spreadsheet. However if I meet a solid problem I will admit it is not viable yet. The same if the trials reveal a new problem.