Autonomous Car Progress

[DanCar]

It's funny that you keep claiming that Waymo is too safe when we have no idea how safe it is yet.

I've spoken to people at Waymo. My opinion is born from those conversations. When I talk to Waymo folks I always hear the same thing, over the past 8 years. They will not take chances. I've stated several times, I don't know how they can be successful without taking chances. I hope I'm wrong, but I think Elon is going to eat their lunch 5 years down the road.

I'm a little confused about why you think public pressure is necessary when you live in the center of autonomous vehicle testing and where Waymo and Cruise both have permits for testing without a safety driver.

Public pressure will be needed for Tesla to remove the nags. Both from a regulatory perspective and a lawsuit perspective. I'm not concerned about Waymo nor Cruise, because Waymo is overly safety conscience and Cruise hasn't proven much to me. Putting public pressure will result in more deaths from Tesla, but hopefully fewer deaths overall. Ones we can't see that Tesla saved us from. Hope that clarifies.
It is only a matter of time before some government agency informs Tesla , they can't do what they are doing in the interest of safety. Perhaps government agency is right, if people start to fall asleep and that causes accidents.

 

[diplomat33]

It is only a matter of time before some government agency informs Tesla , they can't do what they are doing in the interest of safety. Perhaps government agency is right, if people start to fall asleep and that causes accidents.

The NHTSA has already put out a statement that they are closely monitoring Tesla's FSD and will take action if they think it is unsafe:

"“NHTSA has been briefed on Tesla’s new feature, which represents an expansion of its existing driver assistance system,” reads the full statement to the media. “The agency will monitor the new technology closely and will not hesitate to take action to protect (the) public against unreasonable risks to safety.”"

And PAVE (Partners for Automated Vehicle Education) put out this statement:

"“Public road testing is a serious responsibility and using untrained consumers to validate beta-level software on public roads is dangerous and inconsistent with existing guidance and industry norms,” Ed Niedermeyer, communications director for PAVE, says. “Moreover, it is extremely important to clarify the line between driver assistance and autonomy. Systems requiring human driver oversight are not self-driving and should not be called self-driving.”"

Tesla's release of new 'self-driving' software closely watched by U.S. regulator

Regulators Are Keeping an Eye on Tesla FSD Beta, Some Aren’t Exactly Pleased

 

[Daniel in SD]

Public pressure will be needed for Tesla to remove the nags. Both from a regulatory perspective and a lawsuit perspective. I'm not concerned about Waymo nor Cruise, because Waymo is overly safety conscience and Cruise hasn't proven much to me. Putting public pressure will result in more deaths from Tesla, but hopefully fewer deaths overall. Ones we can't see that Tesla saved us from. Hope that clarifies.
It is only a matter of time before some government agency informs Tesla , they can't do what they are doing in the interest of safety. Perhaps government agency is right, if people start to fall asleep and that causes accidents.

It doesn't really clarify. A driverless vehicle by definition does not have nags.
If you're saying that we need to sacrifice lives to develop FSD so that we can save lives in the future I strongly disagree.

 

[Daniel in SD]

The NHTSA has already put out a statement that they are closely monitoring Tesla's FSD and will take action if they think it is unsafe:

"“NHTSA has been briefed on Tesla’s new feature, which represents an expansion of its existing driver assistance system,” reads the full statement to the media. “The agency will monitor the new technology closely and will not hesitate to take action to protect (the) public against unreasonable risks to safety.”"

And PAVE (Partners for Automated Vehicle Education) put out this statement:

"“Public road testing is a serious responsibility and using untrained consumers to validate beta-level software on public roads is dangerous and inconsistent with existing guidance and industry norms,” Ed Niedermeyer, communications director for PAVE, says. “Moreover, it is extremely important to clarify the line between driver assistance and autonomy. Systems requiring human driver oversight are not self-driving and should not be called self-driving.”"

Tesla's release of new 'self-driving' software closely watched by U.S. regulator

Regulators Are Keeping an Eye on Tesla FSD Beta, Some Aren’t Exactly Pleased

Hmm... Thinking about changing my prediction to never on my poll (POLL: When will "Quantum Leap" FSD be released to the HW3 fleet?)
My gut tells me there is no way for testing to be safe without well trained safety drivers who are monitored and work under threat of termination. Though I suspect that right now it might be safe enough because no one in their right mind would become complacent. It's actually when it gets "good" that it becomes a bigger problem.
Bring on the disagrees.

 

[diplomat33]

Conflicting report on FSD. One person says phantom braking is solved. The other says the car drives like a drunk driver.

 

[diplomat33]

30 minute unedited drive on FSD Beta:

 

[daniel]

If they only sell their FSD as L2 (supervision required) but it does everything for you, 10000x safer then the average driver.
You'd still be here pushing the goal posts around.

I have not pushed any goal posts. I'm saying the same thing I've said from the start: I want a car I can sleep in the back of while it drives me. Level 2 is not that. As a first step, I've said I'd pay for Level 3 (I could stop watching the road and the car would take responsibility for telling me when I needed to take over. Level 2 is not that.)

"IF" is not an argument. You say: "If they only sell their FSD as L2 (supervision required) but it does everything for you, 10000x safer then the average driver."

First of all, Read what I said above and what I keep saying: I want to sleep while the car drives. By definition Level 2 is NOT that! Second, you've pulled the number of "10000x safer" out of your ass, because we still have no idea how much safer the new software is. It's much too early. Though some of the videos have shown that it is nowhere near ready for level 3, much less Level 4.

Finally, kindly stop telling me what I would do in some hypothetical situation! That's just bullshit! I have made it clear that I hope they succeed and I want them to succeed and when and if they do succeed I will buy the car or pay for the upgrade on mine. Whatever the cost. The videos above are very impressive. But they clearly show the car is not yet ready for the driver to stop paying attention (as Tesla itself says clearly and loudly: THIS DOES NOT MAKE THE CAR SELF-DRIVING!!!)

Sad but true. Tesla won't be allowed to save lives early on.

People hate the idea of robots killing people. So until it has almost no chance of killing, then people will start to put pressure on law makers. At least 5 years away maybe 10.

I disagree. The insurance companies have powerful lobbies. Once a true self-driving car has demonstrated that it is safer than a human driver, insurance companies will demand that regulators authorize its use.

EAP on the highway, used properly, makes my car safer than if I didn't have it. City driving requires much quicker response times and it's far too early to know if the new software makes the cars safer, or is merely a test-bed. We've seen that the car does make mistakes.

I'm very happy that people are getting the new software and will be able to play with it. I want them to. And to have fun, safely. For me, my car is not a toy, it's transportation. I'm not interested in playing with new features. I will buy them when they're ready and when the evidence shows that they make my car safer. I don't think they can get there with the present software, but I sincerely hope I'm wrong, because I want a true self-driving car.

 

[mikes_fsd]

Conflicting report on FSD. One person says phantom braking is solved. The other says the car drives like a drunk driver.

I wonder how much the stance of the car makes a difference.
Just today, there was a post (see quote below) where using the mounting plate of the tri-cam from Model S on a Model X would not allow the cameras to get calibrated properly.

If the pitch of the cameras affected it that much to fail calibration, I wonder if there is a bigger perception difference with the low-slung Model 3 (re phantom braking) when compared to the tallish Model X.

my mistake/screw up in not keeping the existing mounting plate when I swapped them. The salvage car it came from must have been a Model S rather than X, so the pitch was wrong. All now working perfectly with AP2.5 cams all round; recalibrated really quickly after sorting the mount.

 

[Buckminster]

Elon hasn't mentioned that this is feature complete. Thoughts?

 

[S4WRXTTCS]

Conflicting report on FSD. One person says phantom braking is solved. The other says the car drives like a drunk driver.

I can't say I trust @teslaownersSV because don't have a good track record when it comes to being truthful about the performance of the FSDBeta.

They'll say things like Smart Summons is working great while the video they show clearly shows it pausing for an excessive amount of time for no particularly good reason.

Or the car won't be the in the correct lane before an upcoming turn, and they'll totally gloss over it.

Where @brandonee916 on the other hand is perfectly honest about what's happening.

We can get data from both of them, but we have to factor in how each of them handles things. Brandonee will take over if there is any possibility of it screwing up, and the other guy mostly lets it ride though while glossing over what's happening.

 

[DanCar]

New version of PSD (partial self driving) pushed out: https://twitter.com/brandonee916/status/1319940952715218944

 

[DanCar]

Elon hasn't mentioned that this is feature complete. Thoughts?

Awesome. I think we can imagine features that are missing. Will it pass on a two lane way freeway behind a large truck? Potholes?

 

[DanCar]

Although I mentioned that Waymo had this tech 5 years ago, It won't take Tesla 5 years to catch up to where Waymo is today. More like 2.5 years because:

1. Tesla has better ability to collect odd corner cases
2. Advances in A.I. level the playing field. Most machine learning advances are made public.
3. Tesla is more of a risk taker. Tesla is more aggressive at pushing updates.

Some possible Tesla disadvantages:

1. Fewer sensors
2. Smaller team
3. Less experience than Waymo
4. Tesla won't implement Waymo's remote monitoring

I'll guestimate Tesla will start robotaxi service in 2.5 years.

 

[diplomat33]

Although I mentioned that Waymo had this tech 5 years ago, It won't take Tesla 5 years to catch up. More like 2.5 years because:

1. Tesla has better ability to collect odd corner cases
2. Advances in A.I. level the playing field. Most machine learning advances are made public.
3. Tesla is more of a risk taker. Tesla is more aggressive at pushing updates.

I'll guestimate Tesla will start robotaxi service in 2.5 years.

Those are advantages that will help Tesla make fast progress. Tesla may catch up to where Waymo is today in 2.5 years. But Waymo is not standing still. Waymo is a moving target. Waymo will continue to make progress with their FSD too. In 2.5 years, Waymo will have much better FSD. So I think it is harder to say if Tesla can catch up to Waymo in 2.5 years.

Also, as we see in the FSD Beta videos, Tesla's hardware is clearly good enough for the FSD that Tesla is doing now. Tesla's FSD is still very early. We don't know if it will be good enough in the future. For example, Tesla might hit a wall where they can't get over 5,000 miles per disengagement on the current hardware and they need better sensors or more sensors. This could slow down Tesla in the future.

 

[Grifon]

The NHTSA has already put out a statement that they are closely monitoring Tesla's FSD and will take action if they think it is unsafe:

"“NHTSA has been briefed on Tesla’s new feature, which represents an expansion of its existing driver assistance system,” reads the full statement to the media. “The agency will monitor the new technology closely and will not hesitate to take action to protect (the) public against unreasonable risks to safety.”"

And PAVE (Partners for Automated Vehicle Education) put out this statement:

"“Public road testing is a serious responsibility and using untrained consumers to validate beta-level software on public roads is dangerous and inconsistent with existing guidance and industry norms,” Ed Niedermeyer, communications director for PAVE, says. “Moreover, it is extremely important to clarify the line between driver assistance and autonomy. Systems requiring human driver oversight are not self-driving and should not be called self-driving.”"

Tesla's release of new 'self-driving' software closely watched by U.S. regulator

Regulators Are Keeping an Eye on Tesla FSD Beta, Some Aren’t Exactly Pleased

 

[Grifon]

Tesla’s rollout is to city streets only. That’s explicitly stated in the release notes and the behavior of the software itself. That can be seen in the FSD demos where the car moves from city street to interstate highway, the UI flips into the ‘old’ NOA view and behavior. I don’t see how NHTSA has any jurisdiction currently.

 

[diplomat33]

Tesla’s rollout is to city streets only. That’s explicitly stated in the release notes and the behavior of the software itself. That can be seen in the FSD demos where the car moves from city street to interstate highway, the UI flips into the ‘old’ NOA view and behavior. I don’t see how NHTSA has any jurisdiction currently.

I imagine the NTSB has jurisdiction.

 

[mikes_fsd]

Although I mentioned that Waymo had this tech 5 years ago, It won't take Tesla 5 years to catch up

Those are advantages that will help Tesla make fast progress. Tesla may catch up to where Waymo is today in 2.5 years.

You guys are living with a bad delusion.
Tesla did not have to catch up to Wyamo because Waymo today --- by their own design -- cannot do what Tesla is able to do.

• Random city
  • came back to add this, nothing stops https://twitter.com/Brandonee916 from driving to Reno NV using the exact same release.
• Random Tesla car
• Random Roads

You can argue until you are blue in the face that they could "technically do it", but they publically state they DO NOT and WILL NOT do it.

I am starting the timer to measure how long it takes Waymo/Cruise at al to catch up to what Tesla has in 2020!
and mark!

 

[diplomat33]

they publically state they DO NOT and WILL NOT do it.
View attachment 602022

Correct. Waymo will not do FSD without HD maps. Because doing FSD without HD maps would not meet Waymo's high standard of safety and reliability! How many times do I need to explain this to you?

Tesla's approach is to start with crappy FSD and make it better, hoping that they can eventually achieve safe and reliable FSD without HD maps. Tesla's FSD requires constant supervision and tries to drive into incoming traffic.

Waymo's approach is to use HD maps so that they can start from the get go with safe and reliable FSD. Waymo's FSD with HD maps, does not require driver supervision and is safe that it can operate without any driver in the seat.

Personally, I like FSD cars that don't require driver supervision, that are safe and reliable, and don't try to drive into incoming traffic. But that's just me.

I am starting the timer to measure how long it takes Waymo/Cruise at al to catch up to what Tesla has in 2020!
and mark!

LOL. Keep telling yourself that if it makes you feel better.

 

[mikes_fsd]

The NHTSA has already put out a statement that they are closely monitoring Tesla's FSD and will take action if they think it is unsafe:

"“NHTSA has been briefed on Tesla’s new feature, which represents an expansion of its existing driver assistance system,” reads the full statement to the media. “The agency will monitor the new technology closely and will not hesitate to take action to protect (the) public against unreasonable risks to safety.”"

And PAVE (Partners for Automated Vehicle Education) put out this statement:

"“Public road testing is a serious responsibility

I love it how you tried to put PAVE in the same category of authority as NHTSA.

Good try.
PAVE is a steaming garbage pile with Ed Niedermeyer as the mouthpiece!