Prediction: Cross-Country Self-Driving Tests Already Happening

[DaveM410]

Electrek posted photos at the end of April of a Model 3 spotted at a Supercharger near Cincinnati, overnight, under the cover of darkness.... Link

This got me thinking.... Could the much anticipated LA to NY fully self-driving trip that's planned to happen "by the end of 2017" be a massive head-fake?

Could Tesla come out later this year & say, "by the way, about that cross-country trip, we've already completed it, 50 times over (or something like that).
An announcement like that would make world-wide headlines.....

If (something) like that isn't happening, why was there a Model 3 in Ohio in April?

 

[Tam]

... massive head-fake...

I don't see any problem for those who keep practicing and practicing before an official event.

Athletes do all the time.

No matter how many times they did before and how great they did before, what counts is if they can demonstrate what they can do in an official event.

 

[ggr]

Electrek posted photos at the end of April of a Model 3 spotted at a Supercharger near Cincinnati, overnight, under the cover of darkness.... Link

This got me thinking.... Could the much anticipated LA to NY fully self-driving trip that's planned to happen "by the end of 2017" be a massive head-fake?

Could Tesla come out later this year & say, "by the way, about that cross-country trip, we've already completed it, 50 times over (or something like that).
An announcement like that would make world-wide headlines.....

If (something) like that isn't happening, why was there a Model 3 in Ohio in April?

There are supposed to be a bunch of Model 3s in Ohio for the big testing track setup there. It's quite possible that they're already testing the cross country road trip, but if so they'd be doing it in a Model S or X. They know enough not to debug two things at the same time.

 

[lunitiks]

Could the much anticipated LA to NY fully self-driving trip that's planned to happen "by the end of 2017" be a massive head-fake?

Massive, yes. They're actually done now. Reason we're seing Teslas pop up east and west coast is because of the Boring tunnels. A vast network right under our feet

 

[johnnyS]

judging by how well my autopilot II works, I would be surprised if those model III autonomous cars do not have serious body damage.

 

[Tam]

...doing it in a Model S or X. They know enough not to debug two things at the same time.

I think theoretically, it doesn't matter which model is being trained for Full Self-Driving Capability. It's fleet learning so the learning is all transferable from one car to another within the fleet.

 

[Bladerskb]

One thing Elon does is hype and lie in order to self cars and stock. I mean the hyping is ridiculous. Its obvious from anyone that they are behind in the sdc tech.

So what does Elon do? well hype it up and say they have level 5 capability. claim a better version of their ap1 will come in december and claim that significant FSD features will be released definitely 3-6 months afterwards.

We are well past december and there are no EAP features, nor do we have full ap1 parity, just more promises.

We are also on the verge of the 6th month and there are no FSD features.

At this point if you believe anything elon says and buy a car/stock because of it, you deserve to be hoodwinked!

The cross country drive has been done hundreds of times. its MEANLESS!

 

[lunitiks]

So what does Elon do? well hype it up and say they have level 5 capability

There is no roadway motor vehicle in existence that has SAE level 5 capability.

 

[HankLloydRight]

The cross country drive has been done hundreds of times. its MEANLESS!

Not on 100% auto-pilot.

I think it was Elon who said you'd be able to summon your Tesla in Los Angeles to come pick you up in NYC...with the automatic snake superchargers and FSD.

Yeah, I don't buy it either.

I do believe they will try a 100% AP self-driving cross country trip soon enough, and even call it a success (er, remember "MISSION ACCOMPLISHED"!??! ).. but it will have lots of exceptions and disclaimers that it will be meaningless.

 

[Bladerskb]

Not on 100% auto-pilot.
I do believe they will try a 100% AP self-driving cross country trip soon enough, and even call it a success (er, remember "MISSION ACCOMPLISHED"!??! ).. but it will have lots of exceptions and disclaimers that it will be meaningless.

Actually yes, 100% highway autopilot including changing lanes, ramps off/on.
Delphi comes to mind.

The 10-15 miles of urban street which consists only of taking over to navigate to rest area, gas station refill, etc is mean-less when comparing highway autonomy.

Also the actually cross country is mean-less there's no difference between 3k, 1k or 500 miles drives. Making the genius book of records is mean less when it comes to actual consumer product.

 

[HankLloydRight]

Actually yes, 100% highway autopilot including changing lanes, ramps off/on.
Delphi comes to mind..

Link? There's a car that has driven 100% autonomously across the country, including charging/fueling stops? Oh, you said "highway" autopilot. BFD. That's not the target.

The 10-15 miles of urban street which consists only of taking over to navigate to rest area, gas station refill, etc is mean-less when comparing highway autonomy.

And the disclaimers are already starting. 100% cross-country *except* the 10 to 15 miles of surface streets. That's NOT what Elon proposed. He said you can summon your car from LA to pick you up in New York, including charging stops and surface streets. That's not going to happen, disclaimer free, for a very, very, very long time.

Also the actually cross country is mean-less there's no difference between 3k, 1k or 500 miles drives. Making the genius book of records is mean less when it comes to actual consumer product.

Of course it means something...otherwise it wouldn't be a goal that everyone is reaching for. Are you an economist? You can't just assume away the actual mileage(distance) and say that 500 miles is as good as 3000.

 

[Bladerskb]

Link? There's a car that has driven 100% autonomously across the country, including charging/fueling stops? Oh, you said "highway" autopilot. BFD. That's not the target.

That is the target, the PR of elon doesn't want you to see that. it wants you to see think its demoing fully self driving car while its only demoing highway autonomy.

Read Elon words closely and you will spot the deception.

"It should be able to go and be very good assuming once you enter a highway to go anywhere on the highway system in a given country"

Notice how he also said "assuming once you enter a Highway" and "highway system".
What they are really demoing is highway autonomy and not urban street autonomy.
Its not even a discussion, its outright fact. There is only 7.6 miles of urban street.

charging/fueling is pointless for a sdc tech.

What tesla is trying to do has already been done multiple times.

Delphi self-driving car goes coast-to-coast, autonomously - ExtremeTech

Of course it means something...otherwise it wouldn't be a goal that everyone is reaching for. Are you an economist? You can't just assume away the actual mileage(distance) and say that 500 miles is as good as 3000.

No one is trying to do that. anyone trying to do that is trying to do a PR stunt. Many companies have already mastered highway autonomy which tesla is still struggling. Google in 2012, Audi in 2015, etc. Everyone now is testing almost 95% on non-highway roads.

For example google (waymo) in 5 or so cities have done 3 million miles and out of those miles only a fraction percentage were on highways (probably less than 5%).

 

[M0DEL³]

And the disclaimers are already starting. 100% cross-country *except* the 10 to 15 miles of surface streets.

I would consider it a huge step forward it AP2 were to be able to handle all the crazy detours I dealt with on my last extended road drive: (closed highway resulting in driving into split oncoming lanes on opposite side of median, staggered lanes, flagmen batching vehicles on long single lanes, etc.) If I had to bet there's quite a few more disclaimers yet to come. For now I suspect the best move for AP is to just recognize construction zones and refuse to engage within them.

Also, I'm curious if anyone has any experiences to report on how AP2 reacts to lane splitting motorcyclists?

 

[jbcarioca]

...

If (something) like that isn't happening, why was there a Model 3 in Ohio in April?

The track where Tesla is doing testing for Model 3 is in Ohio, less than two hours northeast of Cincinnati Supercharger at Blue Ash. There were dozens of Model 3's there a few weeks ago, and may still be, although I have not checked during the last couple of weeks.

 

[Derek Kessler]

The 10-15 miles of urban street which consists only of taking over to navigate to rest area, gas station refill, etc is mean-less when comparing highway autonomy.

Actually, those 10-15 miles are FAR more meaningful than highway. Urban driving with all of its distractions, surprises, and complexity is far more important and dangerous than highway. Highways are controlled environments in comparison to city streets — being able to handle stop lights, cross traffic, turn lanes, etc. will be the true test of FSD. Frankly, I'll be much more impressed by a Tesla handling driving around Los Angeles for three hours than cruising down the highway to the next Supercharger exit.

 

[Derek Kessler]

The track where Tesla is doing testing for Model 3 is in Ohio, less than two hours northeast of Cincinnati Supercharger at Blue Ash. There were dozens of Model 3's there a few weeks ago, and may still be, although I have not checked during the last couple of weeks.

Dozens of Model 3s at the testing track, not the Supercharger. Right?

 

[Reciprocity]

That is the target, the PR of elon doesn't want you to see that. it wants you to see think its demoing fully self driving car while its only demoing highway autonomy.

Read Elon words closely and you will spot the deception.

"It should be able to go and be very good assuming once you enter a highway to go anywhere on the highway system in a given country"

Notice how he also said "assuming once you enter a Highway" and "highway system".
What they are really demoing is highway autonomy and not urban street autonomy.
Its not even a discussion, its outright fact. There is only 7.6 miles of urban street.

charging/fueling is pointless for a sdc tech.

What tesla is trying to do has already been done multiple times.

Delphi self-driving car goes coast-to-coast, autonomously - ExtremeTech




No one is trying to do that. anyone trying to do that is trying to do a PR stunt. Many companies have already mastered highway autonomy which tesla is still struggling. Google in 2012, Audi in 2015, etc. Everyone now is testing almost 95% on non-highway roads.

For example google (waymo) in 5 or so cities have done 3 million miles and out of those miles only a fraction percentage were on highways (probably less than 5%).

Haha.. this makes me laugh:

"Delphi estimates that the electronics on the car could cost as little as $5,000 in just a few years. Variants of what’s on the testbed Audi will appear on production Audis in a couple and a radar/camera subset will be on the much-anticipated Volvo XC90."

As little as $5000 in a couple of years. Holy crap whats it cost today? $25000 and then entire trunk of the car and 20% its range to power the whole mess. Good luck with that.

 

[Bladerskb]

Actually, those 10-15 miles are FAR more meaningful than highway. Urban driving with all of its distractions, surprises, and complexity is far more important and dangerous than highway. Highways are controlled environments in comparison to city streets — being able to handle stop lights, cross traffic, turn lanes, etc. will be the true test of FSD. Frankly, I'll be much more impressed by a Tesla handling driving around Los Angeles for three hours than cruising down the highway to the next Supercharger exit.

We are saying the same thing.

 

[Bladerskb]

Actually, those 10-15 miles are FAR more meaningful than highway. Urban driving with all of its distractions, surprises, and complexity is far more important and dangerous than highway. Highways are controlled environments in comparison to city streets — being able to handle stop lights, cross traffic, turn lanes, etc. will be the true test of FSD. Frankly, I'll be much more impressed by a Tesla handling driving around Los Angeles for three hours than cruising down the highway to the next Supercharger exit.

We are saying the same thing.

 

[Bladerskb]

Haha.. this makes me laugh:

"Delphi estimates that the electronics on the car could cost as little as $5,000 in just a few years. Variants of what’s on the testbed Audi will appear on production Audis in a couple and a radar/camera subset will be on the much-anticipated Volvo XC90."

As little as $5000 in a couple of years. Holy crap whats it cost today? $25000 and then entire trunk of the car and 20% its range to power the whole mess. Good luck with that.

You do realize that was 2 years ago?
Plus you think tesla is the only one who has acces to nvidia? Isnt that naive?

Look at audi zfas that went from taking up the entire trunk while in development to miniaturized to the size of a tablet in production this autumn (housing both mobileye eyeq3 and nvidia X3). With watt usage far less than nvidia drive px 2.

How can u not grasp the fact that all development hardware regardless of what type of tech are huge before their production counterparts?

 

[Reciprocity]

You do realize that was 2 years ago?
Plus you think tesla is the only one who has acces to nvidia? How naive are you?

Look at audi zfas that went from taking up the entire trunk while in development to miniaturized to the size of a tablet in production this autumn (housing both mobileye eyeq3 and nvidia X3). With watt usage far less than nvidia drive px 2.

The fact you cant grasp that all development hardware regardless are huge before their production counterparts is laughable.

Ok, so now they can sell you a $5k system. Awesome. Go Audi.

Edit: Hardware would be sized to match the software. nVidia sells the hardware and Yes Audi can buy it, but if the software is bad and requires 10x more processing power then nVidia wont work.

 

[Bladerskb]

Ok, so now they can sell you a $5k system. Awesome. Go Audi.

Edit: Hardware would be sized to match the software. nVidia sells the hardware and Yes Audi can buy it, but if the software is bad and requires 10x more processing power then nVidia wont work.

That link is about delphi who will use a intel SOC with nvidia chip and mobileye eyeq4 combined.

Audi however already uses nvidia and has their system in a production car.

Secondly no bad software requires 10x drive px2. What you seem not to understand is that nissan for example were using multiple titan x. The equvalient of drive px2 is four of them.

If u had 4 titan x thats 650 watts each.
Thats 2600 watts combined and thats withot cpu and other processors power consumption.

You laugh at others for having computer racks and high wattage but thats normal.

Again audi went from a trunk of computers to a tablet size board. It has nothing to do with software.


From this with Thousands of watts

To this with less than 30 watts