Elon Musk: When it comes to autonomous driving, data is everything

Tesla auotpilot

Vehicle autonomy is developing on a parallel course with electrification. Tesla is at the forefront of this trend with its Autopilot features, keeping it a nose ahead of any potential competition from other automakers.

Every day, more Autopilot-driven miles are added to the body of data that the company is amassing. Tesla said in April that Model S and X owners have driven over 47 million miles on Autopilot since it became available last October.

As Musk explained during Tesla’s Q1 2016 earnings call, “Data is everything,” not only for improving the Autopilot technology, but also to convince regulators, and the public, that autonomous features make automobiles safer.

“Regulators are going to want to see a very large amount of data – maybe billions of miles – showing that the car is unequivocally safer in autonomous mode…in a wide range of circumstances, in countries all around the world,” said Musk.

While Musk believes autonomous technology can reduce the death toll from auto accidents, he doesn’t see himself ever arguing that humans should be required to yield the wheel to machines. “Freedom is important. If people want to drive, they should be allowed to drive, but the autonomous safety system should be there.”


Source: Tesla

  • brian_gilbert

    I have been advocating going completely driverless as the technology is already more than adequate. In making my case a cost for the vehicles was needed and Tesla is the only company offerring that at $35,000 for the model 3. The vehicles would be bought in bulk and hired by the trip by operators like UBER. Because the vehicles would be available for use 24 hours a day, centrally maintained by the operator, driverless, electric, and smoothly driven their life would be indefinite. They would only be replaced when outdated. So the cost per mile would be far lower than currently. The Tesla interior could be changed easily enough to something like the rear of a limousine. Once accident risk was proven to be negligible seat belts would not be required. The incar computer controls are probably suitable already only the central control system being needed. That would be pretty much like existing train control systems with vehicles travelling at comstant speed in lanes and switching as required.

    • John Trotter

      Brian Gilbert writes a well thought comment, it’s clear that
      he has spent time reflecting on autopilot driving. But his comments did not address some easily
      imagined concerns.

      Consider the transition from autopilot driving to human
      driving. Since this can be needed at any time, a responsible driver will need
      to be quickly available at all
      times. Where is the benefit ? Such a transition introduces delay into the
      process and increases risk.

      Consider the matters of cleanliness or damage in the real
      world. It will be some time before systems
      deal well with mud, bird poop, or even a wet leaf, let alone rock damage or electronic
      failure anywhere along the line. Until
      then the system’s reaction would likely be to call for the driver to take over. See above.

      Accidents will still happen, that’s their nature. Lawyers will love these cases, arguing over
      who was (or should have been) in control.
      Insurance costs will rise.

      After an accident of any
      size (say your kid smacks the car putting the lawnmower away) and repairs are
      made, who will certify the original performance of the car ? This will be expensive both in dollars and
      confidence. If a second accident should
      occur after a repair, see the note about lawyers above.

      Autopilot landings are not used for commercial
      airplanes. Why not ? That’s a situation far better defined than
      driving on public roads.

      I think we are in the honeymoon phase of an alluring concept
      – we loved the idea of flying cars for a long time. My dear mother-in-law simply wanted to die
      before the metric system became law. I’m
      beginning to feel like I wouldn’t mind if autopilot driving waited until I was

      • brian_gilbert

        No transition problem. I envisage completely driverless operation. You call up the vehicle as with a taxi service, tell it where to go and it takes you there. All vehicles are driverless not just the cars/taxis. The control is simpler than if mixed with human driven vehicles. The vehicle just has to keep to a lane like a railway train and switch lanes when the control system tells it to do so.
        – Vehicles are routed empty for cleaning by the operating company as required. A tow behicle could collect it if it breaks down. Heathrow Personal Rapid Transit very rarely has a breakdown. With all vehicles driverless and maintained by the operator and electric they leave neglible material on the roads. The vehicle has many detection systems as widely reported and will stop if there is a problem. The operator is alerted and will deal with it meanwhile rerouting the vehicle that was obstructed.
        – Insurance is no more a problem for the user any more than when they use a train or a taxi.
        – Accidents will be very few compared to the 1 million deaths a year currently world-wide. There have been no deaths or injuries that I know of since Morgantown, West Virginia Personal Rapid Transit went into operation on 1975,
        – A spreadsheet suggests Singapore could gain US$35 Billion a year by going completely driverless. Off the cuff the same logic suggests US$1 Quadrillion per year for the world.
        – Sadly few countries will face things like finding reemploymemt for drivers so they will only take it seriously when a country like Singapore, Hong Kong or China leads the way.
        – I have been searching the internet for technical problems with completely driverless zones for quite some time and am still looking.

        • John Trotter

          Thanks for the pointers to the Heathrow and Morgantown systems. I wasn’t aware of them and looked them up. I would only point out that they are both ‘guided rail’ systems, orders of magnitude less complex than driving on public roads or city traffic.
          Your comments focus on the future when all this has settled out. Do you have thoughts about the transition period between now and then ?

          • brian_gilbert

            Thanks for the constructive reply.
            – Two other Personal Rapid Transit Systems in operation are Masdar in Abu Dhabi since 2010 and Suncheon in South Korea since 2015. Both use dedicated guideways which makes them costly but they demonstate many important features are workable. There vehicles are also costly because of low numbers and earlier technology.

            – The problems of driving on public roads do not arise as being completely driverless a vehicles cans travel at constant speed in a lane. In the light of trials, barriers against pedestrians can be placed where needed. Control system dictates route of vehicle to prevent congestion roughly doubling capacity of existing roads.

            – Transition Period: Dominated by politics but resistance will crumble when early birds demonstrate the benefits.. 5 to 10 years for Singapore, Taiwan and Hong Kong, 15 for China,, US and UK.

          • Electric Bill

            Only a couple of comments on this topic. Re: seat belts, it need not be either-or. Until all traffic Is autonomous, which is unlikely for the foreseeable future, there will still be the possibility of, say, drunk drivers running red lights and T-boning the car you are in. Seat belts would save your life.

            Seat belts should still be mandatory TO INSTALL, but optional to use.

            Vandals are likely to be a fact of life until the day we manage to forge a Utopian, Stepford world. Until then, security cameras should be installed in all such vehicles and require either positive ID or full-face, hi-res face shot to nix any yen to ruin the interior of the car. If they even so much as leave trash behind, they can be fined, or barred from future use.

          • brian_gilbert

            Thanks for raising those points.

            I have no fundamemtal objection to seat belts and am happy to see sest belts installed initially. I expect a fully driverless system to be installed as I see no solution in sight if mixed with human-driven cars with the resulting deaths, injuries and damage. There is no problem with cost as the annual savings are so large.

            Yes I expect CCTV to be installed in all vehicles as it is in existing Personal Rapid Transit systems such as the one at Heathrow Airport, Terminal 5. I expect most people will be happy to sacrifice their privacy for the protection it will provide as they do on London Buses and trains.

    • Michael B

      I see neither the inevitability nor the benefit of foregoing the seat belt requirement. Sudden braking might still happen (e.g., a child or animal darts in front of the car), and regardless, it is a good reminder that one is in a moving vehicle and it is not inconvenient at all.

      • brian_gilbert

        I agree it should be dependent on trials. At present standing is the commonest method of travel on commuter routes and strangely it is still expected on busses and trains in conditions illegal for animals. . People seem quite comfortable without a seatbelt on umcrowded trains and in planes apart from takeoff, landing and bad weather. Even without seatbelts I would expect deaths and injuries to reduce by well over 90%.