Autonomous Car Progress

[EVNow]

Yes. The fact that the comparison videos only show one drive and don't show the drives where FSD beta required a safety disengagement.

So you assume there are drives that will have safety disengagement (and not getting stuck like Cruise) that they are purposely avoiding - and that is your "proof" ?

I think you owe an apology to CallasEV.

Lets get this straight
- 1 hour of disengagement free video from MobilEye : Fantastic
- 1 hour of disengagement free video from Omar : Shill
- Multiple head-to-head comparisons of Waymo-FSD where Waymo is slower : Fraud

Such brilliant reasoning and logic !!

 

[spacecoin]

Such brilliant reasoning and logic !!

Now you're just trolling. It's a flawed comparison. If you take a photo of a kid jumping on a trampoline you can't present it as proof that they can fly by using it. But you can imply it in marketing and let people dream it. Sounds familiar?

 

[zoomer0056]

Tesla isn't even playing the same

Yup.

 

[diplomat33]

So you assume there are drives that will have safety disengagement (and not getting stuck like Cruise) that they are purposely avoiding - and that is your "proof" ?

It is not an assumption. It is a fact that there are drives where FSD Beta would require a safety disengagement.

I think you owe an apology to CallasEV.

No, I really don't.

Lets get this straight
- 1 hour of disengagement free video from MobilEye : Fantastic
- 1 hour of disengagement free video from Omar : Shill
- Multiple head-to-head comparisons of Waymo-FSD where Waymo is slower : Fraud

Such brilliant reasoning and logic !!

Any 1 hr disengagement free video is great for what it is. They can often be very informative to showcase some of the capabilities of the system. But no video can prove the system is 100% reliable. They can only prove that the system can do that particular route under those particular conditions with zero interventions. And I hold Mobileye and Tesla to that same standard. Mobileye videos do not prove 100% reliability, they only prove that it can do that one route, under those conditions, with zero interventions. The fact that I might share a Mobileye video does not mean that I think it proves Mobileye is better than Tesla. And Omar is definitely a shill but his 1 hr disengagement free videos are great too for what they are. They showcase some good capabilities of FSD Beta. But they don't prove that FSD Beta is 100% reliable.

So to be be clear: I have no problem with zero intervention videos as long as people understand that they cannot prove 100% reliability. And that is my standard for everybody.

And if I tout Waymo as being super reliable, it is not because of a video, it is because Waymo has already proven reliability by being driverless over millions of miles and with safety data. Mobileye and Tesla have not proven reliability since they still require driver supervision.

 

[zoomer0056]

They can only prove that the system can do that particular route under those particular conditions with zero interventions....

... at that particular time ... taking into consideration weather, traffic, time of day, software version and other things I can't pull out of my arse at this time.

 

[EVNow]

No, I really don't.

Yes you do. With zero proof you are slamming a person of fraud.

 

[diplomat33]

Yes you do. With zero proof you are slamming a person of fraud.

Again, you are putting words in my mouth. I am not accusing them of fraud. I never said anything about fraud. I am simply saying the video is not a fair comparison for many of the reasons that I have outlined already. The video is just a snapshot in time. Myy point is simply that you cannot use the video to say that FSD beta is ahead of Waymo since the video is only showing one route under one set of conditions. Just because FSD beta happened to do that one route with zero interventions and faster (since it used highways), does not mean it is better than Waymo in other conditions. At best, you might argue that it was "better" in that one instant, but you don't know how FSD beta would compare to Waymo under different conditions, different routes.

 

[spacecoin]

Lol there's no way spacecoin is some serial entrepreneur programmer, unless he does fart apps

Man you "forgot" to reply to my response regarding camera deficiencies? In a few pages you'll say that no one could explain to you why camera-only HW3/4 autonomy won't work again?

Let's not do that:

OK let's summarise why HW3/4 isn't good enough.

1. Guessing range from a 2D image with only semantic cues is not safe enough, especially when there are few reference objects and at night. Tesla's been running into things (first responders, motor cyclists) at night. It is safer to physically measure the distance, so why wouldn't you?

2. If you travel at 60 mph, and there is smoke or fog on the road, cameras cannot see. If the cameras are blinded by oncoming traffic or the sun, they cannot see. There are too many failure modes for cameras for them to be safe enough to trust with your life.

3. Even if Tesla would add more cameras (in the a-pillars and the front for example) why wouldn't you want to use more sensor modalities to make a 10x safer product for almost zero cost? Tesla won't even bother to add the HW4 radar to 3/Y. Tesla picks lowering cost over improving FSD safety every time.

 

[zoomer0056]

All this back and forth has caused a couple extra synapses firings ... grand Prix of assisted or autonomous driving. Closed course, fake pedestrians, emergency flashing lights, UPLT, multiple routes possible ... Winner based on quickest time, interventions, accidents, fake pedestrian injuries ...

 

[spacecoin]

Your knowledge is way behind. We're not gonna spend the next 5 pages educating you about L4 and how it has nothing to do with smoke or 60mph.

Ok what operational design domain are you talking about then for your L4 in 12 months. Perhaps we just misunderstand each other?

What do you think the Tesla with your kids in the back seat and no one in the drivers seat will be able to do in 12 months? In what weather? On what types of roads? What speeds? NYC? SF? Highway? Rural? Day and night? Heavy rain? Blizzard? Fog?

Educate me sir! I am a quick learner, so probably only a few posts will be required.

 

[powertoold]

Ok what operational design domain are you talking about then for your L4 in 12 months. Perhaps we just misunderstand each other?

What do you think the Tesla with your kids in the back seat and no one in the drivers seat will be able to do in 12 months? In what weather? On what types of roads? What speeds? NYC? SF? Highway?

Educate me sir! I am a quick learner, so probably only a few posts will be required.

Dude, we've been talking about Cruise and Waymo. Cruise is in SF running at a top speed of 30mph.

My whole point was that the bar for L4 is low.

You say stuff like cameras not reliable enough without any examples, just your illogical thoughts.

 

[spacecoin]

Dude, we've been talking about Cruise and Waymo. Cruise is in SF running at a top speed of 30mph.

My whole point was that the bar for L4 is low.

So what will your SAE J3016 Level 4 autonomous robotaxi Tesla be able to do with no one in the driver's seat in 12 months?

You say stuff like cameras not reliable enough without any examples, just your illogical thoughts.

I gave examples but according to you they are a "word salad". Perhaps if you ask a concrete question if there was something that was hard to follow, I can explain it better? Rather than rambling about "fart apps"? Please?

 

[Dewg]

Fun watching you guys argue. I'll toss this experiment in:

Do a city street route near you with FSD Beta, and roll down the speed to limit to 30MPH for the entire drive. Does it perform better, worse, or about the same?

 

[spacecoin]

Do a city street route near you with FSD Beta, and roll down the speed to limit to 30MPH for the entire drive. Does it perform better, worse, or about the same?

Thanks for the thought experiment. Since we're chatting about reliability, I feel the question should be "does it remove the need for a driver" or "can it perform the route 1000 drives in a row" Or even 10 in a row.

"Perform better" is, or can be, functional to me. This discussion is not about "can the car do it sometimes or often".

 

[powertoold]

"can it perform the route 1000 drives in a row during different time of day"

Can fsdb perform a single SF route 1000 times in a row from 12am to 5am (without an at fault accident)? I'm sure it can.

 

[spacecoin]

Can fsdb perform a single SF route 1000 times in a row from 12am to 5am? I'm sure it can.

Bullish! But not true. Ever looked at FSD beta tracker?

 

[powertoold]

Bullish! But not true. Ever looked at FSD beta tracker?

This is what I mean by your illogical posts.

Give us concrete examples why, videos of 11.4.2?

 

[Dewg]

Thanks for the thought experiment. Since we're chatting about reliability, I feel the question should be "does it remove the need for a driver" or "can it perform the route 1000 drives in a row" Or even 10 in a row.

"Perform better" is, or can be, functional to me. This discussion is not about "can the car do it sometimes or often".

Gotta start with a safety driver for the first 1000 drives.

 

[spacecoin]

This is what I mean by your illogical posts.

Give us concrete examples why, videos of 11.4.2?

It's very logical to me since I believe looking at data is more important than me digging up a few videos where beta acts up and you responding with 10 where it doesn't.

According to the tracker there is a disengagement about every 10-15 miles and only 90-something percent of the drives are without a CRITICAL DISENGAGEMENT (FSD breaking the law or would hit things). This number doesn't include other DE:s for lane selection et c or any interventions (such as accelerator taps or speed adjustment).

So, no, you would not get those 1000 drives in a row. You would struggle to get 10 in a row if you drove a 1000 times.

Is that clear and concrete enough?

 

[powertoold]

Is that clear and concrete enough?

No, it's not. Go watch some 11.4.2 videos and show us. You're clearly lacking in knowledge and experience, so you need to get more of it and then come back here, rather than spouting your poor logic.