Elon Musk: Enhanced Autopilot To Be Released In 3 Weeks, Incremental Monthly Releases Beyond That

7 months ago by Steven Loveday 19

A self-driving Model X with new enhanced hardware was demonstrated by Tesla in October - Watch

A self-driving Model X with new enhanced hardware was demonstrated by Tesla – Watch

Despite the fact that all Tesla vehicles produced now have full self-driving hardware, the cars can’t actually drive themselves. Truthfully, while Tesla is still working on the software, the Autopilot features are somewhat “featureless”. Of course consumers are anxious, but Tesla is going to great lengths to assure that the Enhanced Autopilot software is ready, prior to rollout.

Let’s not confuse Enhanced Autopilot with level 5 autonomous “functionality”. Yes, the new cars have level 5 autonomous “capability”, but it will be a year or two before it is fully functional. Not to mention ongoing legislation and a myriad of other possible roadblocks that could delay full autonomy even further. Enhanced Autopilot, on the other hand, is set to release very soon. According to Tesla:

Enhanced Autopilot adds these new capabilities to the Tesla Autopilot driving experience. Your Tesla will match speed to traffic conditions, keep within a lane, automatically change lanes without requiring driver input, transition from one freeway to another, exit the freeway when your destination is near, self-park when near a parking spot and be summoned to and from your garage.

Elon Musk's Autopilot Tweet

Elon Musk’s Autopilot Tweet

Musk confirmed, via Twitter, that Enhanced Autopilot will become available in “about three weeks”, with the system’s advancement and progress happening incrementally in future monthly releases.  Once complete, the software will be updated over-the-air, as will the official level 5 autonomy, when the time comes.

Video (below): Recent Tesla self driving demo with recognition feed:

Features will differ slightly between first and second generation Autopilot vehicles. For instance, second generation Summon will be “Smart Summon” which will be more capable of advanced maneuvers. Autosteer+ for the second generation will be more capable, due to additional cameras. This will prove true of most, if not all, features and incremental updates.

With all of this being said, the goal would be to continue to release the updates and work toward the goal of full autonomy in 2018.  The hope is that federal regulations will keep pace and allow for the 100% integration to coincide with legislation.

Tags: , , , , , , , , , , ,

19 responses to "Elon Musk: Enhanced Autopilot To Be Released In 3 Weeks, Incremental Monthly Releases Beyond That"

  1. John says:

    I’ll be interested when it can pick up my pizza for me!

    1. Trollnonymous says:

      …..and Beer.

    2. Waiting says:

      I may have mentioned this in another post, but why the heck do we have to pay to have the autopilot equipment put on the car if we have no plans to ever activate autopilot? Sure seems like the car could be produced for a lower cost if all that autopilot equipment was eliminated.

      Stop building cars with stuff that the consumer doesn’t want and will never use. What if somebody NEVER wants to use the supercharger system….why would they then have to pay for that equipment being installed whether it’ll be used or not????

      1. Joshua says:

        1) It is less expensive to create an assembly line to produce one car with all of the hardware than it is to custom create hundreds of variations of the same car at the skeleton and hardware levels.
        2) If you don’t provide every car with the supercharger capabilities, it will be virtually impossible to work towards introducing supercharger infrastructure across the nation.
        3) If you don’t want the added safety features of their level 2 autonomy or supercharger capabilities for future cross-country travel, why would you consider a Tesla? If you only want to travel within the city and without additional safety from autonomy, there are other electric vehicles that are much less expensive that can satisfy those needs.

  2. VazzedUp says:

    Wild, tho interesting to see the pauses at points where no issue existed. System got spooked by some runners and a few junctions where it pulled out then stopped once on the new road. So much code and tech to integrate.

  3. HeisenberghtCRAZY says:

    When I stated that autonomous cars will be available on 8.8.2018 most people told me crazy… Now it seems within reach. Call me a proud predictor!

  4. Pushmi-Pullyu says:

    The article says:

    “Yes, the new cars have level 5 autonomous ‘capability’, but it will be a year or two before it is fully functional.”

    Well, that’s what Tesla claims. I don’t believe it. I don’t believe it’s possible to achieve real-world, operational Level 5 autonomy (fully autonomous) without a 360° rooftop-mounted active scanner, either lidar or radar. Just look at the video Tesla released recently, showing all the objects well off to the side of the road which were initially and incorrectly identified as being in the path of the car (link below).

    This simply isn’t good enough for full autonomy. In my opinion, the question isn’t whether or not that’s true; the only question is how long Tesla will continue to deny the obvious: That a sensor suite having only front-facing radar, video cameras, and extremely short-range ultrasonic sensors, is far from adequate for a self-driving car to safely operate on all public roads.

    http://insideevs.com/tesla-releases-self-driving-demonstration-with-recognition-feed-video/

    1. Four Electrics says:

      +1!

    2. Roy_H says:

      Humans have been doing pretty good with just video for some years now. Multiple cameras pointed to cover all 360° is just as good as LIDAR and better in identifying objects IMHO.

      1. Pushmi-Pullyu says:

        Humans have a very sophisticated visual processing center in our brains; the product of billions of years of evolution. Human visual acuity is better than almost any other animal. So far as I know, the only animals with better visual acuity are raptors (hawks, eagles, ospreys, and a few other birds of prey).

        Contrariwise, the undependability of optical object recognition software is well documented. Sadly, one demonstration of this was the one and only fatal accident in a Tesla car confirmed to have been controlled by Autopilot/AutoSteer. Scanning lidar certainly would not have confused the side of a tractor-trailer painted white with a “brightly lit sky”!

        1. bogdan says:

          Humans have a much too sophisticated visual processing center in our brains. That’s what makes it pretty slow.

          Human can concentrate only on one thing, the computer can concentrate on million things simultaneously.

          A computer with a brilliant software can drive the car much better then the best driver in the world.

          1. mr. M says:

            The best system today can not identify correctly if a jogger is moving on the sidewalk or on the street.

            I can.

            The best system today can not interpret mimics of people standing close to the street to interpret if they want to cross it soon.

            I can.

            1. bogdan says:

              Well, that’s the reason why Tesla isn’t releasing level 5 autonomous capability yet.

          2. Pushmi-Pullyu says:

            bogdan said:

            “A computer with a brilliant software can drive the car much better then the best driver in the world.”

            Not if it has to depend on trying to interpret video images from cameras, it can’t. There is a good reason why Google and Lyft use scanning lidar for their self-driving cars.

            As has been said, active sensors are much better for object scanning than passive sensors like video cameras. Tesla is trying to do more with less. I understand Tesla’s desire to save money, and perhaps to avoid using a rotating scanner which might need frequent replacement. But in my opinion they’re just kidding themselves about being able to use video cameras as the primary sensor system.

            At any rate, I’ve made my prediction, and we’ll have to wait and see whether it proves to be right or not.

        2. Heisenberght says:

          On the other hand scanning lidar could identify a flying piece of waste (think aluminium crisps bag) as a truck or something else, is hackable from outside, has interference issues with other lidars and many other things…
          Again, as stated before I am not at all anti lidar. But I appreciate the Tesla approach to use cheap teach like cameras which is available NOW instead of waiting 2-3 years until lidar on a chip or something alike will be cheap enough to be integrated into every car they produce.

          Most likely they will integrate lidar when the cost of it will allow them. Meanwhile Tesla is taking over the job to increase reliability of camera based object recognition and path prediction.

          What we see in the videos is a really early version, we may discuss on whether or not they should make it publicly availible but waiting for cheap lidar is not an option imo.

          In 2018 (to be exact 8.8.2018) we will see a tesla with a quite sophisticated camera suite (let’s say 8 on upper and lower edges, 12 to the front and back (short medium and long range 2 each for 3d…)) some of these maybe infrared.

          Additionally they will have lidar and radar.

          The crappy ultrasonic sensors will stay…

          And Tesla will have microphones all around!

          The options to extend the sensor suite are near limitless… Think wavelengths, lenses and so on… Will we see zoom lenses? Maybe… For the time given the sensor suite seems to be a nice compromise between cost and benefit. As prices fall that suite will become incredibly complex in some years.

          What Tesla is gaining right now is knowledge about sensor data fusion. And they are gaining data data data data… And big huge data!

          That data won’t be lost once they implement lidar. And it surely won’t be worthless!

          Btw in 2025 (the year when VW will build a long range EV…) all new Teslas will be able to smell! They will thereby avoid driving through a burning forest while all lidar equipped cars will burn down due to misclassification of the flames!

    3. ItsNotAboutTheMoney says:

      The hardware suite is sufficient for Level 4 autonomy, but not Level 5.

      If you don’t have Level 5 hardware, you can’t operate the car in conditions that a human couldn’t operate it in, so cameras are sufficient.

      Autonomous vehicles are going to need 360 cameras anyway.

      1. Pushmi-Pullyu says:

        Let’s look at what Level 5 autonomy actually means, by comparison to Level 4:

        [quote]
        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
        Level 4: This is what is meant by “fully autonomous.” Level 4 vehicles are “designed to perform all safety-critical driving functions and monitor roadway conditions for an entire trip.” However, it’s important to note that this is limited to the “operational design domain (ODD)” of the vehicle—meaning it does not cover every driving scenario.

        Level 5: This refers to a fully-autonomous system that expects the vehicle’s performance to equal that of a human driver, in every driving scenario—including extreme environments like dirt roads that are unlikely to be navigated by driverless vehicles in the near future.
        ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
        [unquote]

        Nothing there suggests the car should be able to do anything a human driver can’t.

        source:
        http://www.techrepublic.com/article/autonomous-driving-levels-0-to-5-understanding-the-differences/

      2. Heisenberght says:

        Sorry to be rude but I think that you overestimate your capacity to guesstimate what can be done with that sensor data.

        Unless you work for Tesla and have deep insight into the raw data which is produced by the low amount of cars with the new sensor suite there is no value in your statement.

        In fact even if you had that data and insight you (as well as the people working on that data) can not predict what can be achieved with that suite. It can be level 4 or 5.

        It all depends on how that data is used.

        It also depends on what additional data will be gained due to vehicle to vehicle communication which could be achieved with the current cars to at least a certain degree.

        It also depends on the work done by all Tesla drivers. A stop sign classification which is false today won’t be in 2 weeks… Fleet learning… Tesla has quite a lot of free workers…

        Unless I lend you my crystal ball you will have to wait and see!

  5. Peter says:

    “Enhanced Autopilot To Be Released In 3 Weeks”.

    Just in time for Christmas. Will it be wrapped nicely and put under the tree, with a card signed by Elon?