4 ms·
Let's take a scenario. This seems to happen all the time on the freeway. You're cruising down the freeway at 75 MPH. All of a sudden, traffic comes to a scre
by blackrock 8y ago
Let's take a scenario. This seems to happen all the time on the freeway.
You're cruising down the freeway at 75 MPH. All of a sudden, traffic comes to a screeching halt.
There are 3 cars in this scenario. The first car is the one that slams on his brakes, and comes to a stop.
The second car, cannot stop in time, and he swerves out of the way, and safely gets into the adjacent lane. Close call, but he drives away safely.
The third car, is the Tesla, with the driver on Autopilot, and browsing Facebook on his iPhone. So, he is inattentive.
The whole sequence plays out in less than 5 seconds.
In this scenario, the third car, the Tesla, will not recognize the situation that the second car swerved out of the way. So the Tesla will collide into the first car at high speeds. The velocity of the impact kills all occupants inside the first car, and the third car, the Tesla. So, if both cars are full of occupants, then maybe up to 10 people are now dead.
The problem is that the Tesla does not have situational awareness. It cannot read the current road condition. It cannot see through the tinted windows of the 2nd car, to see that the first car has slammed on his brakes.
So, the accident severity of the Tesla, is much greater, than compared to a scenario where a human was driving the third car.
In this scenario, I think the human might be able to see the situation, and attempt to avoid it. Or to at least mitigate the severity of it. Like, he would slam on his brakes, or steer out of the way, or something. He might still collide with the first car, but probably at a much lower velocity, that all occupants of both cars can still survive.
It just seems that the severity of the accident with current robotic cars, just seems greater, than compared to when a human is the driver.
- chc 8y agoNow let's add a fourth car, a Toyota without TACC, whose driver is also browsing Facebook on his phone. This driver also does not recognize the situation and will collide into the first car at a high speed. Is this worse than the third car? No, probably not. In short, the assumption that a human driver without Autopilot is paying attention and will swerve out of the way is not a valid one. People drive distracted in traditional cars all the time. BTW, from what I have heard, Teslas might actually be better at seeing through other cars than human drivers are, since radar can bounce in ways visible light can't.
- jonhendry18 8y agoA Toyota driver probably wouldn't put that much faith in a mere Toyota. Tesla drivers who are high on Elon's farts put that much faith in their cars. "BTW, from what I have heard, Teslas might actually be better at seeing through other cars than human drivers are, since radar can bounce in ways visible light can't." Or maybe not, since a Tesla couldn't even see a stopped fire truck.
- chc 8y agoPeople who drive normal cars are on their phones while driving all the time. There's no "probably" about it — I know lots of people who have had their cars hit by inattentive drivers. I myself got rear-ended by a guy who was honest enough to straight-up admit that he didn't see me stop because he was messaging on his phone. Also, I'm not sure what connection you see between recognizing a stationary fire truck and seeing two cars ahead with radar. They seem like pretty different tasks.
- JoshCole 8y agoA quick google search puts the average number of deaths due to car crashes at 3,287 per day [1]. Another one gives me a list of four deaths attributed to autonomous vehicles [2]. Not four per day. Just four. Ever. Memory leads me to recall that statistical reports indicating that vehicles with assistive features like autopilot have a better than traditional safety record [3]. I especially disagree with your thought experiment, because I've watched YouTube videos of situations in which Tesla's autopilot avoided accidents both through stopping and through swerving out of the way [4]. I don't agree with your claim that robotic cars tend to have more severe accidents than traditional cars. [1]: https://www.google.com/search?q=average+number+of+deaths+via+car+accident+per+day&ie=utf-8&oe=utf-8&client=firefox-b-1-ab https://www.google.com/search?q=average+number+of+deaths+via... [2]: https://en.wikipedia.org/wiki/List_of_autonomous_car_fatalities https://en.wikipedia.org/wiki/List_of_autonomous_car_fatalit... [3]: http://www.iihs.org/iihs/news/desktopnews/stay-within-the-lines-lane-departure-warning-blind-spot-detection-help-drivers-avoid-trouble http://www.iihs.org/iihs/news/desktopnews/stay-within-the-li... [4]: https://youtu.be/FrJ2uPRRtz0?t=46 https://youtu.be/FrJ2uPRRtz0?t=46 The short seller thesis for TESLA is that since the company is one which is valued based on its story, not its underlying fundamentals, it follows that its price will be volatile in response to negative press (accidents, missed projections, etc). There is a clear path from this observation to making money via negative press. If you're interested in hearing more about how the world is a lot better than you thought it was, but the truth of our situation is distorted by competing forces, here are some cool TED talks [5][6]. [5]: https://www.youtube.com/watch?v=yCm9Ng0bbEQ https://www.youtube.com/watch?v=yCm9Ng0bbEQ [6]: https://www.ted.com/talks/hans_rosling_shows_the_best_stats_you_ve_ever_seen https://www.ted.com/talks/hans_rosling_shows_the_best_stats_...
- dirktheman 8y agoI'm sorry, but I think you're wrong. You make it sound as if all Teslas are suddenly as dumb as their phone-wielding drivers. The Tesla recognizes a stopped car on the freeway much, much earlier than a human driver. The forward collision warning system is always attentive, never blinks. It won't panic slam the brakes, but will brake with exactly the required force. You really are overestimating the ability of human drivers here. If human drivers are so excellent in reading road conditions and anticipating to traffic, why do we have so many accidents? The severity of accidents with robotic cars is a magnitude lower, simply because they always pay attention. It's just that fatalities with autonomous vehicles are newsworthy, whereas fatalities with human drivers aren't. It's a classic example of the Baader-Meinhof phenonemon. Just because you read about it more, doesn't mean it actually happens more. I'm not saying that Teslas are 100% safe or that Tesla shouldn't investigate why these accidents happened. I'm also not saying that there aren't edge cases where a human driver could have prevented an accident where the Tesla would fail. But statistically speaking, autonomous vehicles with all their sensors are a lot safer than human drivers.
- blackrock 8y agoLook, don't get me wrong. I am all for autonomous cars. In fact, I dislike some human drivers so much, that I hope everyone is forced to drive an autonomous car. You know, the idiot drivers that drive erratically, are distracted, swerves in traffic, tail gates, speeds in excess compared to average traffic speeds, etc. These people should get their license revoked. Driving is a burden for these people, and they are better off taking the subway. Or an Uber. But they still need to get to their destination for work, for play, or for their errands. What I am stating, is that the current generation of robotic cars, are inherently unsafe. They need more situational awareness, such as the ability to see all traffic on the road. Including the car in front of the car in front of it. This can possibly be handled by the Vehicle-to-vehicle communication standard that the DOT is proposing. The other scenario might be to make illegal, all non-robotic cars from driving on the freeways and highways. So, all roads that allow you to travel at high velocity, must be a robotic car. But this presents another problem. If you take away people's ability to drive, then you also take away, their ability to get experience with driving at high speeds. So, in the future, you end up raising a whole new generation of people, that don't know how to drive properly at high speeds. Basically, until we get to a level of technology where these cars are fully safe, then the edge cases where accidents and collisions will happen, may end up being more critical with a robotic car, than with a human driver. The danger, lies in the fact, that these robotic cars, will lull the human driver, into a sense of complacency. Where they think that 99.90% of scenarios will keep them safe. And maybe it might. But what about that 0.10% scenario where it won't. And not only is it dangerous for them, then it is also even more dangerous for someone else. And when the robotic car gets into a situation where it needs human assistance, then it is an extreme edge case, where no human can possibly react safely in time. The human driver would never have gotten themselves into that situation in the first place. But even with all that, I am still rooting for robotic cars to succeed, and to safely drive us all one day. And to safely handle that 0.10% scenario.