Testing Tesla Autopilot Summon Mode To See If It Will Detect A Small Child – Video

1 year ago by Steven Loveday 20

There have been many conflicting tests related to the autopilot mode. Obviously there are so many factors making it hard to achieve a “controlled” experiment.

Update: Video embed/playback has been disabled by author – click here to go direct to the YouTube video.

While hitting a child or any human being, for that matter, would be devastating, hitting anything (a bucket, a cone, a trashcan, a small animal, etc.) is not something that is “preferred”. In the video, every test is done is summon mode, without a driver in the vehicle.

All tests were done over a very short distance at extremely low speed. This technology is young and it will take much time to test and perfect it. Drivers must understand that currently autopilot and autonomous modes are only to “assist” the driver that is engaged and responsive and that summon mode requires your attention too.

Video Description:

Tesla Autopilot/Summon: Will It Detect a Child? Or turn them into pulp?

The Tesla On Autopilot Using Summons Mode Does Not Hit The Child

The Tesla On Autopilot Using Summons Mode Does Not Hit The Child

After some fuss in the news about claims a Model S Auto Pilot did not see a child, I decided to do some simple tests to see just what will stop a autopilot or summoned Model S vehicle. Tests should reflect the same for the Model X as well, as it uses the same hardware.

Findings were pretty straight forward. First test using Vinny’s Lucky (Don’t Ask…) Stuffed Duck did not trigger the car to stop. I attribute this to lack of mass of the duck. With so much stuffing, the duck did not have enough “Mass” to reflect back any sonar or ultrasonic from the car.

Test 2, the 5gallon bucket. This in fact did stop the car using the parking sensors. It did not trigger the autopilot sensor.

Third, ME!!! I, somewhat nervously, used myself as a test subject as well. With me, the car saw me in every instance.

Fourth, THE GENO! I had him stand slightly to the side of the car for safety. I also stood within arms reach. I did not risk testing the autopilot sensor as that would require him being directly in front of the moving vehicle, and he would not survive getting run over (Heck, raccoon didn’t even survive last summer!). The car did in fact see Geno.

Hat tip to sven!

 

Tags: , , , ,

20 responses to "Testing Tesla Autopilot Summon Mode To See If It Will Detect A Small Child – Video"

  1. What? No pet Dog or Cat to test it on?
    As it was supposedly upgraded so you have to have a steady grip on your phone and stops moving if you release from touching the phone screen button, and the purpose was essentially to retreave the car from tight or narrow garages, is there still concerns that the car is responsible if you run over a cat?

  2. MDEV says:

    If you have a small child DO NOT USED smart!!!!

  3. Bobby says:

    Lol loved the video

  4. Nix says:

    What does Tesla use to test their system? I’m sure they don’t use children either.

    1. storky says:

      BABY BUSTER

    2. Bone says:

      Perhaps it would make a bit too vulgar video, but some suitably sized animal carcass would be a great and safe test object. It should be simply hanged or supported at suitable height. Carcass has flesh and bones just like human body. AFAIK Tesla doesn’t use infrared sensors (why not?) so the temperature of the test object makes no difference.

  5. This is a rare Model S because it has (what look like) the aero wheels, I think?

  6. kubel says:

    Video was pulled by the owner.

    1. Pushmi-Pullyu says:

      It can still be seen at the YouTube website. Just click on the link at the bottom right corner of the video frame.

    2. Will says:

      That’s not what it says ๐Ÿ˜›

  7. Pushmi-Pullyu says:

    Quoting the audio in this video:

    “…the thing about the [stuffed toy] duck is that it has low mass, and it’s also absorbent, which means it will absorb sound and other radio-style signals as well…”

    Oh, for a facepalm emoticon…

    It would be nice if the guy doing this video had at least a junior-high school understanding of physics.

    Let’s review some basic science:

    Radar works on reflectivity of radio waves. Radio is one frequency band of electromagnetic radiation, along with ordinary light, as well as infrared and ultraviolet light, and more exotic forms of radiation such as x-rays and gamma rays.

    Contrariwise, sound is pressure waves in a solid, liquid or gas medium, and has absolutely nothing whatsoever to do with electromagnetic radiation.

    * * * * *

    Under the test conditions under consideration here, the ability of either radar or ultrasonic sensors to sense an object is dependent on three factors:

    1. The object’s reflectivity to radar or sound

    2. The object’s height

    3. The object’s width

    The tests shown in the video are so far removed from proper scientific testing that it’s laughable. For instance, Tesla says that the sensors may not detect a low object, and this is due to the placement of the sensors on the front of the car. If the object is too low, it will be below the angle at which the sensors can detect it.

    The guy actually does mention the height factor several times, but at no point does he make any attempt to actually test this variable by placing the same object at different heights in front of the car.

    It’s also sadly laughable that he keeps mentioning “mass”. Mass has nothing to do with it. I rather suspect that if you suspended a giant sheet of aluminum foil in front of the car, that either the radar or the ultrasonic sensors would detect it just fine, due to the hard surface and the high reflectivity.

    Bjorn _____ did a video test where the Model S’s front radar failed to detect a giant sheet of thin styrofoam propped up in front of the car. Driving down the road at a good speed, the Model X ran right thru it, scattering pieces of styrofoam in its wake. Again, that’s due to lack of reflectivity. Styrofoam is composed largely of air, so there’s very little for radar to bounce off.

    There was also a Consumer Reports video which showed the car’s sensor failing to detect a bicycle. The bike was parked in front of the car, parallel to it… presenting the narrowest possible profile to the car’s sensors. Tesla warns that the car may fail to detect low or narrow objects, so it seems the bicycle is too narrow to be sensed.

    Now, the guy does have a point about the soft, fuzzy texture of the stuffed toy duck absorbing sound, and therefore being difficult for the ultrasonic detectors to detect. But that’s about the only comment he made that was both correct and relevant.

    * * * * *

    A proper test of the sensors would involve a series of objects, different groups with different reflective properties, with objects in each group having different heights and widths.

    But the bottom line here is: Don’t lay your baby or toddler on the pavement directly in front of the car and then use Summon to move the car toward the child. Duh!

    1. Pushmi-Pullyu says:

      Oops, premature posting error.

      Edit: That should be “Bjรธrn Nyland”, rather than “Bjorn _____”.

    2. theflew says:

      I agree with the things you’ve said, but at the same time imagine the lawsuit someone would be laying at the door of GM, Ford, Toyota, BMW, Mercedes or Honda if the same thing was discovered. That’s why they don’t introduce beta features in their cars that are able to control, throttle, braking and steering. It’s also something Tesla will have to be very careful with concerning the Model 3. You’re talking about a very different buying purchasing a $35k car versus a $75k car.

  8. Bone says:

    But what about people wearing very fluffy outfits? Is Model S a safety risk to people dressed up as ducks?

    1. sven says:

      The Model S is probably a safety risk to wear a Gore-Tex coats. ๐Ÿ˜‰

      1. sven says:

        It’s also a safety risk to people wearing puffy shirts. ๐Ÿ˜‰

        1. Jay Cole says:

          /respect for the puffy shirt reference

  9. IDK says:

    I thought this video was going to have an EPIC ending!

    1. Pushmi-Pullyu says:

      The wildly overdramatic title, “Autopilot vs. Children”, turned out to be quite an overstatement, didn’t it? The only child in this video was offscreen… briefly heard, never seen.

      Apparently the guy making this video thinks that a large stuffed animal is a suitable “stunt double” for a small child. ๐Ÿ˜‰

  10. Delta says:

    I would keep all microwave ovens, power tools, lawn mowers, remote controls, powered garage doors, dry ice, helium balloons, scissors and all other sharp objects away from this guy!