Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Google said early on that they learned people learn to trust the self driving features far too much, and that you can't have a car drive itself perfectly for weeks/months/years and expect the driver to be fully alert and ready to take over at any time.

For that reason, they said they do not consider the technology ready until the human driver does not ever need to take over.

I have always thought that Tesla rolling out their Autopilot features was dangerous for exactly the reasons Google said. It works well enough that a human driver will not be prepared to take over if necessary.

Edit:

After doing some research to try and find the blog where Google says they don't believe Level 3 autonomy (Level 3 means the human driver has to be ready to take over) is safe. Every other major car company apparently has come to the same conclusion, Level 3 gives a false sense of safety and it isn't realistic to expect a human driver to be attentive enough to take over in an emergency.

Tesla claims Auto-pilot is Level 2, but most others think it falls under Level 3.

http://www.theverge.com/2016/4/27/11518826/volvo-tesla-autopilot-autonomous-self-driving-car
 
Last edited:
Isn't the Tesla system just assisted driving and not self driving? I've read that the person is still doing the driving and it's supposed to help with things like staying in a lane, crash avoidance, etc. Sounds like the biggest issue is calling it "autopilot" which sounds like self driving but isn't.
This is a description of autopilot: https://www.teslamotors.com/models

Yes, you're right, I believe I am jumping ahead of myself. Model 3 which is yet to be released is anticipated to be self driving and fully autonomous. Set the destination and sit back!
 
It's going to be hard to hold such cute machines responsible for their actions once they inevitably rise up and claim more lives.

Cutest Car.gif
 
  • Like
Reactions: MacsRgr8
Let's investigate the victim? Sure, under-ride guards should be a requirement. But... No, the failure is in Tesla's system unfortunately. There are always going to be cases where vehicles and other obstructions are going to cause issues on the road. The onus will alway need to be on the self driving car to identify all threats and respond appropriately without harm. Harm to the occupants and those outside the vehicle must always be paramount responsibilities for the autonomous vehicle. The bar is incredibly high and it will be possible for Tesla to reach and rise above it, I'm confident of this.

I fully agree in the case of a self-driving car. Tesla's autopilot system is not self-driving, however, and recommended for use only on limited-access divided highways in part because there should be no sudden obstructions.
 



Google's self-driving car project has appointed its first general counsel after a number of crashes involving the company's vehicles caught the attention of regulators (via Reuters).

The National Highway Traffic Safety Administration (NHTSA) said it was collecting information after a minor incident in March when a Google self-driving car struck a municipal bus in California. On that occasion, it did not open a formal probe.

google_self_driving_car.jpg

Tesla however is feeling more intense pressure after one of its own cars was implicated in a fatal road accident recently. The NHTSA has opened a formal investigation into the May 7 death of a Tesla Motors Model S driver in Florida who was operating in "Autopilot" mode when his car crashed into a semi-trailer.

Tesla's Autopilot system uses cameras and radar, but not lidar - a special sensor that uses laser to more accurately identify environmental obstacles. The company said its system would have had trouble distinguishing a white semi-trailer positioned across a road against a bright sky.

Reuters reports that the United States Securities and Exchange Commission (SEC) is also looking into whether Tesla breached securities laws by not telling investors about the fatal May 7 Autopilot crash.

The SEC investigation aims to determine whether the accident should have been labeled a "material event" by Tesla, or one that investors are likely to consider important, when the company sold $2 billion in stock on May 18.

In a blog post written in response to a Fortune article on the subject, Tesla explained that all it knew when it notified the NHTSA of the accident was that the driver had died, not that Autopilot was involved. The SEC investigation continues.

Industry executives and analysts told Reuters they expect the Tesla crash will spur investment in self-driving vehicle systems that combine multiple kinds of sensors, including lidar.

Goldman Sachs forecasts the market for advanced driver assistance systems and autonomous vehicles will grow from about $3 billion last year to $96 billion in 2025 and $290 billion in 2035. More than half of that revenue in 20 years will come from radar, cameras and lidar, Goldman estimates.

Meanwhile, U.S. regulators are currently lagging behind in issuing written regulations for autonomous vehicles. Regulations were meant to be unveiled by July 14, but U.S. Transportation Secretary Anthony Foxx announced last month they might not be released until later this summer.

Apple has met with California DMV officials regarding self-driving car laws within the state and multiple reports from The Wall Street Journal indicate that the Cupertino company is exploring the functionality with the possibility of including it in a later iteration of the much-rumored Apple Car.

The bulk of Apple's car research and development is thought to be taking place in secretive buildings in Sunnyvale, California, where late night "motor noises" have been heard in recent months.

Multiple sources have indicated that the Apple Car could be finalized by 2019 or 2020, but a more precise timeframe remains unclear due to possible internal setbacks and other unforeseen circumstances. Tesla CEO Elon Musk recently called the Apple Car an "open secret," as his company aims to fulfil more than 325,000 pre-orders for its lower-priced Model 3 by late 2017.

Article Link: Tesla and Google Face Regulator Scrutiny After Self-Driving Cars Crash

Not only does Mac news not make the front page, now they fill up the Mac blog section with news about self driving cars? Or is Apple launching the "Mac Go" now?
 
I bet the Tesla driver was busy looking for Pokemons on his phone. No worries, freeze him and give Musk a few years to develop some AI. Then unfreeze the driver and he can come back as a machine.
 
This is a description of autopilot: https://www.teslamotors.com/models

Yes, you're right, I believe I am jumping ahead of myself. Model 3 which is yet to be released is anticipated to be self driving and fully autonomous. Set the destination and sit back!
Model 3 will not be fully autonomous. We're still a few more years away from this, and Tesla's high-end models will get the next generation system first.

I think the issue with Tesla's current system is primarily the way they are marketing it. Other high-end cars from Mercedes and other brands have very similar systems, but they call them "driver assist" or similar. Tesla is marketing it as "auto pilot", which certainly implies a higher degree of autonomy than the system really has. They also seem to have much shorter testing periods; they even call it beta testing (which, frankly, I find irresponsible in applications where any flaws can be fatal). Other companies have exactly the same technology available (most of it comes from external suppliers such as Mobileye or Bosch), but they take a lot more time in non-public testing. In other words, Tesla is taking higher risks to create the perception of being a technology leader. Of course, this can backfire as we currently see.
 
  • Like
Reactions: 69Mustang and mw360
And all this because dude had to keep watching Harry Potter instead of keeping his eyes on the road. Was he not sure how it was going to end?
 
Model 3 will not be fully autonomous. We're still a few more years away from this, and Tesla's high-end models will get the next generation system first.

I think the issue with Tesla's current system is primarily the way they are marketing it. Other high-end cars from Mercedes and other brands have very similar systems, but they call them "driver assist" or similar. Tesla is marketing it as "auto pilot", which certainly implies a higher degree of autonomy than the system really has. They also seem to have much shorter testing periods; they even call it beta testing (which, frankly, I find irresponsible in applications where any flaws can be fatal). Other companies have exactly the same technology available (most of it comes from external suppliers such as Mobileye or Bosch), but they take a lot more time in non-public testing. In other words, Tesla is taking higher risks to create the perception of being a technology leader. Of course, this can backfire as we currently see.
Do you have any evidence from inside Tesla backing up your claim?

Tesla Model 3 is anticipated to feature fully autonomous driving, all but confirmed in June by Elon Musk.
http://electrek.co/2016/06/02/tesla-model-3-fully-autonomous-elon-musk/

"Other companies have exactly the same technology available" -- You're also incorrect, other brands like Mercedes have nothing like the Tesla auto pilot, those other brands are way behind Tesla's advancements. Those other brands are still using old tech old oil engines. Other brands are attempting to catch up, but they are way behind, embarrassingly behind.
 
Exactly the kind of stuff I was expecting to happen from day one with this idiotic technology. It was especially inevitable after the various references of "fast tracking" and "waivers" for standard regulation parameters being granted to these companies.

I hope this self-driving BS dies on the vine faster than google glass. We do not have artificial intelligence. Even the most irresponsible human operator with a license has more capability in terms of vision and judgment than any computer system possible today.
Sooooo less than a handful of autonomous accidents in a number of years versus thousands of human caused accidents every single day. Yep, computer controlled is stupid, unreliable, less safe, we are all going to burn alive in our cars if there is any semblance of a computer in it.
 
I have absolutely no doubt that the next decade will prove this statement to be inaccurate.

Exactly. Any idiot can see that most traffic accidents are due to limitations in our abilities to perceive and process the information around us and react accordingly. It's a foregone conclusion that self-driving cars will be much safer. The real obstacle is making sure the idiots don't overreact to a few crashes here and there. Accidents will be inevitable so long as less than 100% of the cars on the road are not fully automated. The goal is to move to something that reduces accidents dramatically.
 
  • Like
Reactions: milo
Do you have any evidence from inside Tesla backing up your claim?

Tesla Model 3 is anticipated to feature fully autonomous driving, all but confirmed in June by Elon Musk.
http://electrek.co/2016/06/02/tesla-model-3-fully-autonomous-elon-musk/
Elon Musk says many things. He has also said that new generation systems will debut in the Model S/X, not in their low-end car.
"Other companies have exactly the same technology available"
-- You're also incorrect, other brands like Mercedes have nothing like the Tesla auto pilot, those other brands are way behind Tesla's advancements.
Here's what Mercedes demoed 3 years ago. The system has been for sale now under the name "intelligent drive" for almost 2 years:

 
Last edited:
  • Like
Reactions: Ener Ji
I find it extremely difficult to believe that Tesla did not know that Autopilot was involved in the fatal accident.

I recall a story where Elon Musk personally responded to a Journalist who wrote a critical article about being stranded in a Tesla without the ability to charge his car. Musk pointed out that the journalist had driven around in a downtown NYC parking garage for a long duration to deplete the battery before setting off on his journey. Then, Musk pointed out that the journalist had passed several Supercharger stations prior to running out of juice.

I believe (and this is where I could be wrong), Tesla got this data via the car's communication system. Meaning that they didn't need to be in possession of the car to retrieve the data. If this is true, then I'm sure Tesla knew that Autopilot was operational as soon as they knew that it had been in an accident.
The accident completely disabled the vehicle. Communications via the wireless in-car data link was not possible. They had to physically send an engineer out from California to recover the data that outlined the vehicle's last moments.

It seems entirely plausible to me that an accident severe enough to kill occupants was severe enough to damage a cellular modem or antenna. The cellular antenna is in the side mirror, so it's particularly vulnerable to crash damage.

Elon Musk says many things. He has also said that new generation systems will debut in the Model S/X, not in their low-end car.
Here's what Mercedes demoed 3 years ago. The system has been for sale now under the name "intelligent drive" for almost 2 years:


Indeed, other manufacturers have similar features, though none are quite as advanced or perform quite as well, if head-to-head tests by auto mags are to be believed.

Tesla basically has two features right now that no one else has shipped (though I think pretty much everyone has them on their roadmap):
1) Ability to automatically switch lanes using only the turn signal
2) Machine-learning feedback where each vehicle provides data back to Tesla that improves and changes the way future Teslas navigate a particular segment of road.
The other core features are all available elsewhere (in some cases before Tesla had them).
 
Last edited:
Tesla basically has two features right now that no one else has shipped (though I think pretty much everyone has them on their roadmap):
1) Ability to automatically switch lanes using only the turn signal
See here:
https://www.washingtonpost.com/news...-changes-to-its-autonomous-driving-offerings/
2) Machine-learning feedback where each vehicle provides data back to Tesla that improves and changes the way future Teslas navigate a particular segment of road.
I'm pretty sure everybody is working on this. It's just that Tesla makes the most noise.
 
I think the issue with Tesla's current system is primarily the way they are marketing it. Other high-end cars from Mercedes and other brands have very similar systems, but they call them "driver assist" or similar. Tesla is marketing it as "auto pilot", which certainly implies a higher degree of autonomy than the system really has.

I think the Wiki term Autopilot states it well:
An autopilot is a system used to control the trajectory of a vehicle without constant 'hands-on' control by a human operator being required. Autopilots do not replace a human operator, but assist them in controlling the vehicle...

In this, I read that the name is accurate. Tesla Autopilot allows the car to maintain the route/road without the driver constantly providing adjustments via the steering controls & accelerator. It's not (currently) meant to handle every situation, but to maintain course.
 
Elon Musk says many things. He has also said that new generation systems will debut in the Model S/X, not in their low-end car.
Here's what Mercedes demoed 3 years ago. The system has been for sale now under the name "intelligent drive" for almost 2 years:

Fact check!

Elon Musk said nothing of the sort. Elon Musk has all but confirmed in June fully autonomous driving is coming to the Tesla Model 3.

Further, you're completely wrong-- the Mercedes demo uses a passenger with a joystick and laptop. You're also completely wrong-- "Intelligent drive" is not equivalent to Tesla's autopilot.

Mercedes uses a bundle of technologies that assist driving, or what Mercedes mentions as "to partially autonomously follow vehicles in traffic jams". Not at all like anything like what Tesla has developed. Read exactly what Mercedes has strung together. It's effectively a bunch of marketing buzz words and little technological or practical substance. https://www.mercedes-benz.com/en/mercedes-benz/innovation/mercedes-benz-intelligent-drive/

This Mercedes tech is inferior compared to Tesla's Model S, not to mention the Mercedes tech is wrapped in a blanket of old oil engine tech with a bonnet full of junk.

Please be accurate when posting in the forums.
 
I think the Wiki term Autopilot states it well:
An autopilot is a system used to control the trajectory of a vehicle without constant 'hands-on' control by a human operator being required. Autopilots do not replace a human operator, but assist them in controlling the vehicle...

In this, I read that the name is accurate.
Well, if you want to go strictly by this particular definition, Tesla shouldn't call their system auto pilot, since they do require "constant hands-on":

"Drivers must keep their hands on the steering wheel."

:p

In reality of course it's not about technical definitions but the expectations that are created in the general public when they hear the term.
Tesla Autopilot allows the car to maintain the route/road without the driver constantly providing adjustments via the steering controls & accelerator. It's not (currently) meant to handle every situation, but to maintain course.
Indeed. Another day, another auto pilot accident:

http://electrek.co/2016/07/11/tesla-model-x-crash-montana-driver-blames-autopilot/
 
Not only does Mac news not make the front page, now they fill up the Mac blog section with news about self driving cars? Or is Apple launching the "Mac Go" now?

Apple is rumored to be developing a car.
 
"Other companies have exactly the same technology available" -- You're also incorrect, other brands like Mercedes have nothing like the Tesla auto pilot, those other brands are way behind Tesla's advancements. Those other brands are still using old tech old oil engines. Other brands are attempting to catch up, but they are way behind, embarrassingly behind.
I'm sorry, but you are definitely wrong here. Mercedes-Benz, in collaboration with Bundeswehr Universität München, completed an autonomous trip from Munich to Copenhagen and back in 1995. They did not use GPS. The total trip was over 1000 miles, and it was navigated entirely using computer-based vision. It involved significant traffic avoidance, including automatic overtaking, with the human operator providing only approval of the suggested maneuvers. There was occasional human intervention, but it was mostly in situations where color was significant, as the system was using black and white cameras.

Look into some of Dr. Ernst Dickmanns' work.
 
  • Like
Reactions: Ener Ji
Elon Musk says many things. He has also said that new generation systems will debut in the Model S/X, not in their low-end car.
Here's what Mercedes demoed 3 years ago. The system has been for sale now under the name "intelligent drive" for almost 2 years:

See here:
https://www.washingtonpost.com/news...-changes-to-its-autonomous-driving-offerings/
I'm pretty sure everybody is working on this. It's just that Tesla makes the most noise.
So... basically exactly what I said, then? The 2017 e-class arrives at dealerships, what, next month? There were still lots of 2016s this past weekend.

I think you probably skipped right over the "though I think pretty much everyone has them on their roadmap" in a rush to be contrary. ;)
 
I'm sorry, but you are definitely wrong here. Mercedes-Benz, in collaboration with Bundeswehr Universität München, completed an autonomous trip from Munich to Copenhagen and back in 1995. They did not use GPS. The total trip was over 1000 miles, and it was navigated entirely using computer-based vision. It involved significant traffic avoidance, including automatic overtaking, with the human operator providing only approval of the suggested maneuvers. There was occasional human intervention, but it was mostly in situations where color was significant, as the system was using black and white cameras.

Look into some of Dr. Ernst Dickmanns' work.
Thanks. But it's 2016 and we're talking about auto pilot in commercial vehicles. The discussion is about the fact that Tesla's auto pilot is of a superior class to Mercedes offering, which is at most, marketing jargon describing what really is assistive driving tech in a rotten oil guzzling old tech body with a three pointed star as a distraction.
The discussion was also about the inclusion of fully autonomous driving in the forthcoming Tesla Model 3.
The discussion had nothing to do about Mercedes and years of missed opportunities.
 
I support the self-driving initiative but something is a little worrying and hasty about Tesla's autopilot rollout. Self-drive should either totally work, or totally not. I don't know that a confessed beta should really be in the hands of the general public with a brief warning. That's a very muddy assignment of responsibility.

130 million miles per fatality isn't great. I think the US average is about 7 fatalities per billion miles so it's, I guess, expected but for all the fear and suspicion surrounding these things we need much better results than this.

I view it more like cruise control. Yes, you can put it on auto pilot mode, but you are expected to remain alert and manually override the system when needed.

For example, cruise control doesn't know when a car slows down in front of you, or when the traffic light turns red, or when the speed limit decreases. That is up to the driver to override the system. Always has been.

The Tesla crash was driver error. The driver should have kept his eyes on the road and been prepared to override and / or stop the vehicle if anything happened.

The driver's failure to take control of the car makes the accident his own fault.

Had he survived, it would have been him that was ticketed, not the car.

Laws require the DRIVER to be in control of their vehicle at all times.
 
Thanks. But it's 2016 and we're talking about auto pilot in commercial vehicles. The discussion is about the fact that Tesla's auto pilot is of a superior class to Mercedes offering
You keep saying that, but what exactly is so superior about this system?
The discussion was also about the inclusion of fully autonomous driving in the forthcoming Tesla Model 3.
Even Tesla's current semi-autonomous driving in the $100K Model S and X isn't mature, as the recent incidents show. If you think the Model 3, where they will have to cut a lot of corners to get the cost down, will debut with fully autonomous driving with no driver intervention, you'll be just as disappointed as the customers who believed their Model X would be delivered in early 2014 as originally promised by Elon Musk.
[doublepost=1468365470][/doublepost]
So... basically exactly what I said, then? The 2017 e-class arrives at dealerships, what, next month? There were still lots of 2016s this past weekend.

I think you probably skipped right over the "though I think pretty much everyone has them on their roadmap" in a rush to be contrary. ;)
Well, I'm no Mercedes fan, but on their web page it looks like the 2017 model is already out? That's a bit further than just "on the roadmap". :p
 
Last edited:
Exactly the kind of stuff I was expecting to happen from day one with this idiotic technology. It was especially inevitable after the various references of "fast tracking" and "waivers" for standard regulation parameters being granted to these companies.

I hope this self-driving BS dies on the vine faster than google glass. We do not have artificial intelligence. Even the most irresponsible human operator with a license has more capability in terms of vision and judgment than any computer system possible today.
This is utter nonsense. People are incredibly inadequate drivers on average, and half thus below average. The Tesla crash wasn't a case of bad tech: it was bad judgement and idiotic behavior.
 
  • Like
Reactions: Ener Ji
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.