Crown

Tesla Model S Crashes Into Fire Truck With Autopilot Reportedly Engaged

The Culver City, California, fire brigade shared a couple of photos on social media with a crashed Tesla Model S.

According to their Twitter page, the electric saloon crashed into the truck while doing 65mph (105km/h), on Monday morning, with the Autopilot engaged, as confirmed by the Tesla driver.
 

Luckily, nobody was injured, but the blacked-out Model S will need some TLC before turning heads again, as its entire front end was damaged in the crash. The hood was almost peeled off from the body, and it appears that the front suspension may have been affected as well.
 

On the other hand, the fire truck has minor damages that include a broken taillight, some light scratches, and dents.

“While working a freeway accident this morning, Engine 42 was struck by a Tesla traveling at 65 mph. The driver reports the vehicle was on Autopilot. Amazingly, there were no injuries! Please stay alert while driving!”, the Culver City Fire Department wrote on social media, along with the “distracted driving” hashtag.
 

This isn’t the first Tesla crash with the Autopilot turned on, and probably not the last either, as several Model S and X cars were destroyed while their drivers were relaxing behind the wheel, and at least one owner was fatally injured.
 

Elon Musk’s company continues to advise Tesla drivers to remain vigilant when Autopilot is engaged, as the self-driving technology “is intended for use only with a fully attentive driver”.

If I'm reading that correctly the fire truck was stopped and the Tesla was going 65mph and the driver wasn't injured!  Very impressive accident survival by the driver.  (Accident avoidance is another issue.)

This picture doesn't look like a 65mph impact. The car must have been slowed down before the impact.

And again - that's why the driver should always be aware what's going on. You must be a huge idiot relying 100% on that machine. 

There is no engine in the front so crash structures can be bulked up for extra protection.

Government really need to mandate that for a car to have autopilot function, the driver must give the car some form of input every 30 seconds just so the car knows the driver is still paying attention to the road ahead.

Autopilot works for planes as there aren't other planes within 10 feet and they have tons of space around them, for cars, the reaction time to save the car is next to nothing when there are barriers and other cars within feet of the car.

Whoopsy:

Autopilot works for planes as there aren't other planes within 10 feet some form of input every 30 seconds just so the car knows the driver is still paying attention to the road ahead.

Don't forget in some, Autopilot just hold you in altitude,not taking you to that altitude.

Whoopsy:

There is no engine in the front so crash structures can be bulked up for extra protection.

Government really need to mandate that for a car to have autopilot function, the driver must give the car some form of input every 30 seconds just so the car knows the driver is still paying attention to the road ahead.

Autopilot works for planes as there aren't other planes within 10 feet and they have tons of space around them, for cars, the reaction time to save the car is next to nothing when there are barriers and other cars within feet of the car.

This happens because car manufacturers (especially Tesla but others will follow I'm afraid) are forcing the advertisement, distribution and development of autonomous driving tech.

This technology will not be ready for prime time for another couple of decades (I'm not kidding), unless they start installing some sorts of sensors in the cars(!) and the roads and/or traffic lights/road signs/etc..

I love technology but we just aren't there yet and this is a tech which requires a 99.9% accuracy and only a basically non existent failure rate. Not going to happen for the next couple of decades, sorry to say that.

Cars aren't some smartphones which can run on alpha or beta stage software, the tech needs to be perfect, the software very very sophisticated and bug free (as much as possible). No error margins here, I'm afraid.

--

RC (Germany) - Rennteam Editor Porsche 991.2 Carrera GTS Cabriolet (2018), Porsche Cayenne S Diesel (2017), Audi R8 V10 Plus (2016), Mini JCW (2015)

10 years ago you were using a flip phone. It's not going to take decades for this to happen.

noone1:

10 years ago you were using a flip phone. It's not going to take decades for this to happen.

It does...look at plane tech. Would be much more simple (up, down, keep a steady speed and height, etc.), first attempts were made...so far no self flying planes for a reason. 

We need completely new tech to do that with 99.9% safety, the tech today won't cut it. We need new sensor and camera tech which doesn't exist yet and cost has to be lowered as well, so this tech can be used in every vehicle, maybe even "embedded" in roads, street signs, etc..

We also need a better wireless communications infrastructure.

Autonomous vehicles are the future, indeed but looking at the current tech, I doubt it will take only 10 years.

10 years ago? I had an iPhone.  15 years ago? I was using Windows Mobile touch phones. Tech was basically the same (in theory) but completely different hardware and software.

RC (Germany) - Rennteam Editor Porsche 991.2 Carrera GTS Cabriolet (2018), Porsche Cayenne S Diesel (2017), Audi R8 V10 Plus (2016), Mini JCW (2015)

AI has moved insanely fast in even the last few years. We're in the early inning of an AI (r)evolution and it's going to make big gains really quick IMO.

Sure, 100% human level autonomy might be decades away, but "good enough" autonomy probably isn't.

noone1:

AI has moved insanely fast in even the last few years. We're in the early inning of an AI (r)evolution and it's going to make big gains really quick IMO.

Sure, 100% human level autonomy might be decades away, but "good enough" autonomy probably isn't.

I'm not so sure about it but we'll see. Right now, I don't trust that autonomous stuff too much.

RC (Germany) - Rennteam Editor Porsche 991.2 Carrera GTS Cabriolet (2018), Porsche Cayenne S Diesel (2017), Audi R8 V10 Plus (2016), Mini JCW (2015)

Here is an interesting article on why these crashes against stationary objects happen with these autonomous systems, apparently its got more to do with the limitations of systems based on just RADAR + camera (like Tesla) as opposed to those with LIDAR (as well as other sensors and tech) which are not fully developed nor cost effective for mass production yet. I'm sure it is a question of time but its not just about improving the AI software, the hardware is not here yet either.

https://www.wired.com/story/tesla-autopilot-why-crash-radar/

⇒ Carlos - Porsche 991 Carrera GTS

RC:

noone1:

AI has moved insanely fast in even the last few years. We're in the early inning of an AI (r)evolution and it's going to make big gains really quick IMO.

Sure, 100% human level autonomy might be decades away, but "good enough" autonomy probably isn't.

I'm not so sure about it but we'll see. Right now, I don't trust that autonomous stuff too much.

"Tesla Is Last In the Driverless Vehicle Race, Report Says"

(18 January 2018)

Navigant Research has compiled a new report on 19 companies working on automated driving systems, and investors may be surprised to see Tesla Inc in dead last place in the driverless car race.

Navigant ranked the 19 major companies developing AV technology based on 10 criteria, including vision, market strategy, partnerships, production strategy, technology, product quality and staying power. According to the report, General Motors Co and Waymo, the auto unit of Alphabet, are the top two AV investment opportunities in the market today.

Tesla and Apple are the two biggest laggards in the AV race, according to Navigant's rankings.

Investors are acutely aware of Tesla's production and distribution disadvantages compared to legacy automakers like GM, but Navigant is also highly critical of Tesla's technology.

"The autopilot system on current products has stagnated and, in many respects, regressed since it was first launched in late 2015," Navigant says in the report, according to Ars Technica. "More than one year after launching V2, Autopilot still lacks some of the functionality of the original, and there are many anecdotal reports from owners of unpredictable behavior."

Tesla's first version of Autopilot relied heavily on technology from Mobileye, but Tesla ended its relationship with Mobileye in 2016. Mobileye has since been acquired by Intel Corp.

Navigant analysts aren't the first to question Tesla's driverless technology. in September, Bernstein analyst Max Warburton said Autopilot 2.0 was way behind schedule and did not have the full range of capabilities that the Mobileye-powered Autopilot 1.0 had.

"While Tesla has many advantages, we believe its self-driving capabilities are exaggerated," Warburton said.

Article Link: https://money.usnews.com/investing/stock-market-news/articles/2018-01-18/tesla-inc-tsla-stock

---------------------------

"Navigant Research Leaderboard: Automated Driving Vehicles"

Cleantech Market Intelligence

2017 saw a continuing acceleration of the pace of development in the automated driving sector, as many of the companies involved have shifted from a R&D stance to production engineering. With concerns about driver distraction mounting, automated driving is increasingly seen by many as the best solution to eliminating this dangerous trend on the world’s roadways. The number of automated ride-hailing pilot programs has also increased since it has become increasingly clear that mobility as a service will be the primary means of deploying automated vehicles, particularly in the early years of commercialization.

As the technology for taking the human driver out of the control loop matures, the business landscape continues to shift. The cost and complexity of developing and deploying automated driving technologies have driven numerous new partnerships and acquisitions to help achieve scale more rapidly. OEMs and suppliers have recognized the threats to their traditional business models and are working aggressively to leverage the strengths they have to develop new revenue streams. New entrants into the field are recognizing the challenges of establishing manufacturing and distribution infrastructure and are increasingly forming partnerships with incumbents to leverage the strengths that each bring to the challenge.

This Navigant Research Leaderboard evaluates 19 companies developing automated driving systems. These players are rated on 10 criteria: vision; go-to market strategy; partners; production strategy; technology; sales, marketing, and distribution; product capability; product quality and reliability; product portfolio; and staying power. Using Navigant Research’s proprietary Leaderboard methodology, vendors are profiled, rated, and ranked with the goal of providing an objective assessment of their relative strengths and weaknesses in the development and deployment of automated driving technology.

Top 10 Vendors:

1.  GM

2.  Waymo

3.  Daimler-Bosch

4.  Ford

5.  Volkswagen Group

6.  BMW-Intel-FCA

7.   Aptiv

8.   Renault-Nissan Alliance

9.  Volvo-Autoliv-Ericsson-Zenuity

10.  PSA

Report Link: https://www.navigantresearch.com/research/navigant-research-leaderboard-automated-driving-vehicles

Excellent analysis and one that finally helps to put the superiority myth of Tesla into the rubbish bin.  Tesla is a laggard, not a leader in many automotive systems including product development and manufacture to autonomous driving systems.  One chuckles when Tesla, and its vocal fanbase, promote the its over-the-air upgrade system.  What they fail to mention is Tesla continual delivers Beta or even less complete and tested software to its customer base.  The legacy automakers deliver finished and qualified software and systems.  This is a significant difference between Tesla and its competitors.  

Beta is ideal for cars.

"Hey look, a $150K 2019 G-Class! Oh darn, it has the infotainment from 2012 because the product timing cycle is off by a few months and was developed with the old one. Man, the MBUX is gonna be amazing in 2022! Can't wait for the 2022 G-Class!"

Cars have long life cycles where they don't change at all. When you miss a technology launch, you end up having to wait forever to get it because the old guard would rather ship outdated stuff for 5 years and update at the next facelife/generation. It's dumb.

Carlos from Spain:

Here is an interesting article on why these crashes against stationary objects happen with these autonomous systems, apparently its got more to do with the limitations of systems based on just RADAR + camera (like Tesla) as opposed to those with LIDAR (as well as other sensors and tech) which are not fully developed nor cost effective for mass production yet. I'm sure it is a question of time but its not just about improving the AI software, the hardware is not here yet either.

https://www.wired.com/story/tesla-autopilot-why-crash-radar/

The more important question is, WHY didn't the driver SEE the stationary fire truck and BRAKE! I mean, it's a gigantic big piece of bright red truck.

This is exactly why government need to legislate in driver input at regular intervals for a car to be on 'autonomous' driving. 

Whoopsy:

Carlos from Spain:

Here is an interesting article on why these crashes against stationary objects happen with these autonomous systems, apparently its got more to do with the limitations of systems based on just RADAR + camera (like Tesla) as opposed to those with LIDAR (as well as other sensors and tech) which are not fully developed nor cost effective for mass production yet. I'm sure it is a question of time but its not just about improving the AI software, the hardware is not here yet either.

https://www.wired.com/story/tesla-autopilot-why-crash-radar/

 

The more important question is, WHY didn't the driver SEE the stationary fire truck and BRAKE! I mean, it's a gigantic big piece of bright red truck.

This is exactly why government need to legislate in driver input at regular intervals for a car to be on 'autonomous' driving. 

x2 that is the real problem, the users need to be required to use it correctly and operate it within the limits of the technology, and that means that the current autonomous mode still needs to be continuously supervised by the driver at all times. And for the safety of the occupants and the other drivers on the road that needs to be enforced because there is always bound to be people to dumb, lazy or irresponsible that is not going to, whether its via legislation and/or mechanisms in the tech that force the driver to do so or both.

⇒ Carlos - Porsche 991 Carrera GTS

Carlos from Spain:

Whoopsy:

Carlos from Spain:

Here is an interesting article on why these crashes against stationary objects happen with these autonomous systems, apparently its got more to do with the limitations of systems based on just RADAR + camera (like Tesla) as opposed to those with LIDAR (as well as other sensors and tech) which are not fully developed nor cost effective for mass production yet. I'm sure it is a question of time but its not just about improving the AI software, the hardware is not here yet either.

https://www.wired.com/story/tesla-autopilot-why-crash-radar/

 

The more important question is, WHY didn't the driver SEE the stationary fire truck and BRAKE! I mean, it's a gigantic big piece of bright red truck.

This is exactly why government need to legislate in driver input at regular intervals for a car to be on 'autonomous' driving. 

x2 that is the real problem, the users need to be required to use it correctly and operate it within the limits of the technology, and that means that the current autonomous mode still needs to be continuously supervised by the driver at all times. And for the safety of the occupants and the other drivers on the road that needs to be enforced because there is always bound to be people to dumb, lazy or irresponsible that is not going to, whether its via legislation and/or mechanisms in the tech that force the driver to do so or both.

The Peltzman Effect was created by one of my old professors at UChicago/Booth.  Essentially the Peltzman Effect is people overcompensate for risk mitigation by increasing other risk factors.  The driver was overconfident on the Autopilot’s performance that s/he no longer felt obligated to provide oversight for the system. Autopilot has extremely limited capabilities to handle adverse conditions; this is but one example.  

On some of the more advanced planes, the autopilot is good enough to do a automatic landing, not just flying straight and level in altitude. But the pilots, even when they are not piloting, they are still paying attention and ready grab the yoke at a moment's notice. But flying a plane on autopilot is a lot safer than cars on freeway, they don't have to worry about a suddenly stop plane in mid air or another driver drifting over cause they are texting.

Hell even some boats have autopilot, well more like automatic cruise control, but then again ships and boats have a gigantic ocean to float on, just like planes with the air space.

Still not sure WHY people want to have autopilot in a car. It doesn't work, it's dangerous, it is highly unneeded and it still doesn't work.

Cruise control in car is good enough, they car maintain a certain speed, the driver can take his foot off the gas and rest, but his hands are still in control of the wheel and his eyes are still looking ahead. The foot is always ready to step on the brakes if needed. That should be the end of automation in cars.

Only as dangerous as the people driving them. Autopilot is used by probably 100K Teslas every day and they aren't all crashing. Accidents happen.

Some people can use their photos reasonably safe while driving, others are terribly dangerous and reckless. It is what it is.

noone1:

Only as dangerous as the people driving them. Autopilot is used by probably 100K Teslas every day and they aren't all crashing. Accidents happen.

Some people can use their photos reasonably safe while driving, others are terribly dangerous and reckless. It is what it is.

By the same token some people are perfectly capable of driving at 200+km/hr on the highway, quite a bit here do 300+km/hr every day, why do we have a speed limit? 

Cause of a small minority of drivers who can't handle speed about 100km/hr. 

noone1:

Only as dangerous as the people driving them. Autopilot is used by probably 100K Teslas every day and they aren't all crashing. Accidents happen.

Some people can use their photos reasonably safe while driving, others are terribly dangerous and reckless. It is what it is.

How many other autos, those with active drivers, actually managed to avoid hitting that fire truck?  One is sure that the number is significantly greater than zero.  Don’t classify something as an accident when it was 100% avoidable and avoided by many, many other drivers, most not likely driving a Tesla. 

Most accidents are 100% avoidable. that's why they're called accidents. If they weren't, they wouldn't be accidental. People in non-Teslas get into accidents like this all the time. How? Not paying attention.

I bet 10 stationary cars around the world were rear ended by distracted drivers today alone, all being non-Teslas. Accidents like these are quite common.

Even if the driver was alert, if the autopilot system fails to detect the car in front, it is highly  unlikely that the driver would be able to prevent the accident. That is because autopilots, ACC, etc require some trust from the users of the system for them to let their legs off the brake. So when the user expects it to brake and it doesn't, it might as well be too late. 

These systems will get better but I really don't like the way Tesla are using their customers as their beta testers. It certainly feels that they are always behind schedule so mistakes are more likely to happen in their systems. 

Yup, foot off the pedals is actually a really big deal. When you're on cruise control and suddenly need to brake, it's massively delayed as your foot searches for it. No matter hold long you've been driving or how familiar you are with your car, if you take your foot of the gas pedal you quickly lose all reference of where the pedals are and finding the brakes suddenly is not easy or controlled. In fact, it's a very jarring and nervous experience. It's easy to even first hit the gas before adjusting to the brakes.

Thank you gentlemen!

You all made the argument against self driving car bullet proof!

I couldn't have said it better myself.

There is simply no place for 'autopilot' on highways. 

Suit yourself. You're the minority.

noone1:

Suit yourself. You're the minority.

Aren't you the only one arguing FOR autopilot here if I am not mistaken.

That's some minority I have 

Minority here maybe, but not, you know, in the world of 7B people. In a few years every new car will have these features.

RC:

noone1:

AI has moved insanely fast in even the last few years. We're in the early inning of an AI (r)evolution and it's going to make big gains really quick IMO.

Sure, 100% human level autonomy might be decades away, but "good enough" autonomy probably isn't.

I'm not so sure about it but we'll see. Right now, I don't trust that autonomous stuff too much.

Still think we haven't come very far with AI?

https://giant.gfycat.com/YearlyNewBarb.webm

You guys know I’m the long distance driving champ. You can bet your ass that the last thing I would do during a 20 hour drive is turn on cruise control. Of any type. It is lie taking a sleeping pill. I’m not in favor of any type of auto pilot. 

And planes have been landing on their own for 40 plus years. I was on one in Greece 🇬🇷 when I was 12 landing in Corfu.