Tesla’s Electric Self-Driving Car Euphoria Gets Jolted

An event predicted by many finally occurred: the first fatal crash in an electric self-driving car, on May 7 in Florida. That crash was followed soon by two more non-fatal crashes, one in Pennsylvania on July 1 and another in Montana early Sunday morning. Tesla’s stock fell and the media is having a field day predicting the death of the self-driving car. However, as a big advocate of the self-driving car, here’s my take on it.

It Was Likely the Truck’s Fault

There is nothing really disputable about the basic facts.  A Tesla in self-driving mode is heading eastbound on a Florida highway at an unlawful speed. A westbound truck driver starts to make a left turn in front of the car. The car detected the turning truck as a harmless overhead sign, and took no action, nor did the driver, who was apparently watching a Harry Potter movie on a portable DVD player. The car hit the side of the trailer without braking, the top sheared off, and the rest of the car continued on through two fences, broke a utility pole, and finally came to rest in a front yard. Here’s the diagram of the police report.

tesla-crash-police-report-diagram-superJumbo.jpg.650x0_q70_crop-smart

Here it is on Google Maps. Note how close the car was to a house when it still had enough momentum to break a utility pole.

Tesla Xs

Florida apparently does not have a law like Minnesota does that you forfeit your right-of-way if you are exceeding the speed limit. The Tesla was going straight ahead the truck driver clearly did not yield to the Tesla as required. The same thing could have happened to a person driving the car but not paying full attention. Whether this type of behavior by truck drivers is endemic I don’t know, but as a data point I was involved in a much more minor crash with a truck trailer in the same type of situation, where the truck driver expected me to get out of his way, and I expected him to follow the law.

Also of note, if we held trucks to the same standard as Europe, the result would have been different. European standards require side guards on trailers. Had the trailer been equipped, first of all the Tesla’s sensors would have picked it up and stopped. Secondly, if a crash had occurred, either because the sensors of a self-driving car glitched or because a driver was texting, the top would not have been sheared off, the airbags would have deployed, and if the driver wasn’t able to walk away he’d probably be taken to a hospital instead of the morgue. There are things we can learn from Europe: as another example, amber turn signals are safer, but the U.S. has no interest in requiring them as well as side repeater turn signals (which make it easier for pedestrians to see the intent of the motorist) like the rest of the world does.

One thought I had is why weren’t locations of overhead signs maintained in a database? If the system senses a “sign” that it doesn’t know about, why not sound an alarm to alert the driver, maybe start to slow down. Maybe the false alarms would be unacceptable, but food for thought.

Even In the Extreme Early Days of the Technology, it Appears to Be Relatively Safe

Although this sounds crass, we finally have a single data point; one fatality in 130 million miles of driving. Although we need a lot more data to be sure, it seems to compare favorably to one fatality every 94 million miles in the US or 60 million worldwide. What would be regrettable is if the extreme hype surrounding this incident actually did sour investors and the public on the technology. You can have the type of situation where because of the high profile nature, people get pushed into statistically more dangerous pursuits. 246 non-hijackers were killed in airplanes on 9/11. But one study estimates 1,595 were subsequently killed on the roads because they were too afraid to fly. And whether the TSA, first with their body scanners and now with the outrageous lines here and elsewhere, is doing more harm than good by discouraging people from flying is an interesting question.

The Technology is Just Too Revolutionary to Fail

Just like airplanes many years ago, the technology is just too revolutionary to let if fail. I’ve gone over this in a previous article, but will repeat a summary.  Can’t find an affordable single-family home near your job?  Buy one that you can afford out in Mankato and take a nap or watch movies during your commute to Minneapolis. Want to spend a weekend in Chicago with your car but don’t want to drive? Pack your family in Friday night, take a nap, and arrive Saturday morning with a car to take you anywhere in the city or suburbs  you want to go, then return on Sunday night ready for work the next day and have two full days in Chicago with a car. Too old or otherwise physically unable to safely drive? No need to give up that freedom or rely on the generosity of getting rides from relatives. Admittedly these are all from my “suburbanist” viewpoints, but I’m sure urbanists will have much to say about the transforming effect on the core cities.

The Technology will Only Get Better

One of the most inspiring places I’ve been was Kitty Hawk, NC, the site of the first powered airplane flight. Walking around the obelisk that proclaims “In commemoration of the conquest of the air,” you can see the distance of the first flight.

IMG_1520

In Commemoration of the Conquest of the Air

At 120 feet it was much less than the wingspan of the revered 747, but everything starts someplace. In 1930 there was a Hardy Boys mystery that took place around an airport. The boys muse that “planes nowadays are almost as safe as the car we’re driving in now,” and although they had never been up in one, air mail was becoming a thing and it was obvious to everyone what was to come. Flying really didn’t become accessible to the masses until deregulation and the growth of low-cost carriers, but it came in time. Just like Teslas are now limited to the rich, you can see that won’t always be the case.

Right now, self-driving cars are where airplanes were in 1930s. Very early on, but it’s obvious how transforming it will be and how much potential the technology has to become safer and cheaper. Just like the occasional air crash, no matter how good the technology is, at some point the computer is going to steer a car into a brick wall. But like an air crash it will be front page news, not a daily happening reported on page 5, if at all, like the 116 other car crash deaths every day in the US. Sensors will get better and cheaper, software will improve, and with more self-driving cars on the road there will be fewer humans to do stupid and illegal things.  Where will the monument be to the conquest of the tedium and danger of driving?


Streets.mn is a non-profit and is volunteer run. We rely on your support to keep the servers running. If you value what you read, please consider becoming a member.

, , , ,

4 Responses to Tesla’s Electric Self-Driving Car Euphoria Gets Jolted

  1. Adam Miller
    Adam Miller July 18, 2016 at 3:34 pm #

    I agree with the compelling potential of the technology, but for different reasons. I don’t think there’s a lot of unmet desire for more and longer car trips, although making existing trips less stressful will have a certain appeal.

    But what makes it really compelling is the potential for our road system to be vastly safer and more efficient. Simply following the rules of the road can significantly increase capacity without any infrastructure expansion.

    Of course, it remains to be seen whether driver/passengers will be willing to ride in a vehicle that actually follows the rules.

    • Monte Castleman
      Monte Castleman July 19, 2016 at 8:02 am #

      We’ll see, I guess. Another thing possibility is that some existing gasoline stations will cater to these type of travelers. Upon arrival in Chicagoland after an all-night drive Saturday morning they could offer food and and showers while your car is being charged up in the parking lot.

  2. James Warden
    JamesWarden July 19, 2016 at 12:23 pm #

    A researcher at an autonomous vehicle conference I attended (I seem to recall she was a user interface researcher) argued that the most dangerous period in the adoption of this technology will be when it is nearly autonomous, but not quite fully autonomous. The idea is that users will be inclined during this period to trust the systems with more responsibility than it can safely handle. That is, it’s good enough to convince people it’s safe but not quite yet good enough to be safe in all circumstances.

    Good design could mitigate this by limiting the ability of autonomous measures to take over when circumstances aren’t appropriate, such as when it’s snowing. But many of these circumstances won’t be able to be foreseen, such as when a truck pulls out into the middle of a highway. Google and the traditional auto manufacturers are trying to avoid this uncanny valley of safety by waiting until feature sets are fully vetted by company employees specifically trained to be on the lookout for periods when autonomous functions fail. Tesla, on the other hand, has made its autopilot available to everyday users despite describing its autopilot as a beta. Regardless of whether Tesla is legally culpable in this latest crash, it’s arguable their strategy doesn’t recognize the psychological tendencies that that user interface researcher described. It’s also arguable that it could be bad for the autonomous vehicle technology “brand.”

    Of course, there is still a question about whether the loss of life from inappropriate use of technology outweighs the loss of life from not having the technology. But I could see there being a short-term spike in autonomous vehicles deaths during an interim where:
    a) the technology is first hitting mainstream users not yet accustomed to the inevitable limitations of any technology;
    b) the safest future-gen technology, like vehicle-to-vehicle communication, is not yet widely available; and
    c) there are still numerous conventional vehicles on the road with drivers who are no more predictable than they are today.

    I fully expect this technology to prevail, but we should be ready for some legitimate growing pains.

  3. Nathanael August 8, 2016 at 7:28 am #

    I’m quite sure that it’ll be impossible to make self-driving cars work on roads like the one where this crash happened.

    The road design is bad. Really, really bad.

    That said, the evidence is that the driver of the Tesla was speeding, and is probably largely at fault for his own death. The ‘autopilot’ doesn’t prevent speeding if the driver tells it to speed.