Autonomous Car Progress

[Bladerskb]

It really doesn't matter who goes cross country with no disengagements with level 4 or 3 or what ever. What matters is that whatever tech accomplishes the goals of level 3 or 4 are is in a package that is appropriate for consumers. If all the lidar's and sensors and computers cost more then a fully loaded P100D then it's an all but pointless milestone. The reason why it's more interesting if Tesla does it is because it means it's just a firmware update away from being in customers hands.

No one is actually gunning for that goal other than Tesla. Its a pointless PR play that ofcourse the fraud Tesla/Elon is trying to use to sway PO.

I contend that waymo and Delhi and all the rest have a very rough road ahead to commercialize all that hardware into a package that is cost effective.

Waymo is trying to run a taxi and isn't at all interested in consumer ownable cars.

If you see a self driving demo car with chunks of hardware sticking out of it, it's not something that is commercially viable, yet.

Delphi cars has no sensors sticking out.

http://www.post-gazette.com/image/2...V-Vehicle-Autonomous-Technology-Las-Vegas.JPG

 

[cwerdna]

Tesla has a rough road ahead as well, but its most likely a case of better software and not needing to make major changes to the hardware, outside of upgrading the computer which they have already acknowledged.

So far, there's been little to no evidence Tesla's self-driving software is better than that of say Waymo or Cruise Automation.

As for hardware, who knows if anything else is insufficient? Elon has this whole anti-lidar thing going on which could turn out to be a huge mistake.

 

[Reciprocity]

No one is actually gunning for that goal other than Tesla. Its a pointless PR play that ofcourse the fraud Tesla/Elon is trying to use to sway PO.



Waymo is trying to run a taxi and isn't at all interested in consumer ownable cars.



Delphi cars has no sensors sticking out.

http://www.post-gazette.com/image/2...V-Vehicle-Autonomous-Technology-Las-Vegas.JPG

Are they interested in making money? Because you cant make money with taxi cabs that cost $150k, even if they are driverless.

Driverless only matters if its economical, otherwise you just put a driver in the car.

Tesla's argument is that you will never get there until you solve Vision and that $70,000 lidar is a crutch that will end up causing their failure.

 

[Reciprocity]

So far, there's been little to no evidence Tesla's self-driving software is better than that of say Waymo or Cruise Automation.

As for hardware, who knows if anything else is insufficient? Elon has this whole anti-lidar thing going on which could turn out to be a huge mistake.

This is mostly for people like @Bladerskb or as I call him, the Audi fan boi who lives on the Tesla forum for some odd reason, and others whos sole purpose is really just to bad mouth Tesla and throw around terms like Fraud:

I would say there is clear evidence that Tesla's current progress is worse those mentioned. I dont know if you more technical guys listen to the quarterly conference calls, but Elon explains exactly why the most recent call though I suggest reading the transcript instead. You can either believe him or think he is trying to defraud everyone for $3000 and jeopardizing his entire mission to do so. I tend to believe a guy who has delivered astounding results over the last 10 years including things that many experts had been saying were impossible. I get that everyone is upset that it has not come sooner and Elon is not the greatest at estimating time. I guess there is a chance that Elon himself just sat down one day and said I think we just need 8 cameras and an nvidia GPU with no testing and no due diligence and a month later it was in every car with no clue how it was going to work. I guess that is possible, though I dont know why anyone would think that someone who spent a billion dollars preparing for the Falcon Heavy's first launch would take such risky short cuts. I just dont get what people think that Tesla is getting out of tricking people. I guess Elon could just be delusional and incompetent. But if that was the case, how the heck did Tesla and SpaceX get to this point? Some would have you believe it was all PT Barnum trickery, but I own two Tesla's and I can assure you that its not a trick.

Tesla builds the safest most advanced vehicles on the road today and I am sorry that they have delivered on FSD, but neither has anyone else. Tesla believes that their methodology is the only way it will work and I believe them over anyone on this forum. Not because I dont like you all, but because they are actually risking everything to do it and none of you really risking anything. They do not believe that Lidar is necessary nor does it even add value due to its weaknesses. Again, I have do defer to those who have done more then read articles on the web and frequented Audi forums. No offense to any of you who are Driving Automation experts and AI experts. Instead of debating endlessly about how they cannot succeed, why not think of how they might succeed. You know, just for a second or two. Ask yourself why do they believe they can succeed with this method and take into account the many complicated success they have had in the past.

For those claiming Fraud, I would highly suggest you read up what that actually means. Terms like Ponzi scheme and Fraud are thrown around a lot and I dont think people who use them so lightly know what they actually mean. or worse if they actually know what they mean and are just using those terms stir up crap.

 

[MarcusMaximus]

It really doesn't matter who goes cross country with no disengagements with level 4 or 3 or what ever. What matters is that whatever tech accomplishes the goals of level 3 or 4 are is in a package that is appropriate for consumers. If all the lidar's and sensors and computers cost more then a fully loaded P100D then it's an all but pointless milestone. The reason why it's more interesting if Tesla does it is because it means it's just a firmware update away from being in customers hands. I contend that waymo and Delhi and all the rest have a very rough road ahead to commercialize all that hardware into a package that is cost effective. Tesla has a rough road ahead as well, but its most likely a case of better software and not needing to make major changes to the hardware, outside of upgrading the computer which they have already acknowledged.

If you see a self driving demo car with chunks of hardware sticking out of it, it's not something that is commercially viable, yet.

Thinking a bit more about Apple. You always think software, but Apple is a hardware company and a pretty good one based on the hardware for the iPhone which always dominates the benchmarks. This might be where Apple can make up ground, by doing a better job of shrinking and packaging the hardware required to commercialize self driving tech. I don't know what Google is great at this though they are not novices either. Delhi and the other OEMs certainly have this capability but at what level. It's not like they have been making cutting edge chips and powerful computers with 3D facial recognition VR features that fit in your pocket and run for a full day on a charge. It will be interesting what Apple delivers. I highly doubt they would just jump into self driving without a very good idea of how they would turn it into large profit margins. Because Apple doesn't bother with anything that isn't massively profitable.

A caveat: any ugly looking hardware becomes viable if the cars aren’t meant for personal ownership. If you’re hailing the equivalent of a cab, then who cares what it looks like on the outside?

 

[lunitiks]

You could always hide the rear-facing radar under the bumper, like Model 3

 

[mspohr]

Interesting that Mitsubishi has developed AI for automobile cameras which can sense motion at 100 meters.
AI-Aided Cameras Mean No More Car Mirrors, No More Blind Spots

This would seem to reinforce Elon's position that cameras are sufficient for FSD and Lidar is not necessary.

 

[MarcusMaximus]

I keep coming back to this: we know that vision, alone, is sufficient for driving. Every consumer car to date has been driven using vision alone. There may be questions around whether people can figure out the right type of processing to do, or whether some camera suite has good enough resolution/contrast, etc, but we know it’s possible.

 

[Reciprocity]

A caveat: any ugly looking hardware becomes viable if the cars aren’t meant for personal ownership. If you’re hailing the equivalent of a cab, then who cares what it looks like on the outside?

My comment wasnt about aesthetics. It was about miniaturization which also has nothing to do with aesthetics. It has to do with costs and flexibility as you shouldnt have to design the car around the sensors. The ugly cars with crap hanging off of them means they are very expensive and that moving that tech from where it is today to a final product that is cost effective and mass marketable is a very steep hill to climb. Its similar to the hill that Tesla has to climb with vision. The difference here is that one is a software problem and the other is a hardware problem. Software tends to move faster, but that is not always the case. I think it will be the case here because of the nature of how Neural networks can improve very quickly. They can also fail miserably. I have no idea how Tesla will fare, I just believe them over anyone on this forum.

 

[Bladerskb]

Are they interested in making money? Because you cant make money with taxi cabs that cost $150k, even if they are driverless.

Driverless only matters if its economical, otherwise you just put a driver in the car.

Tesla's argument is that you will never get there until you solve Vision and that $70,000 lidar is a crutch that will end up causing their failure.

Who told you that SDC taxis will cost $150k when deployed in mass?

$70,000? You are still using the same statistics from 10 YEARS AGO!
Waymo already cut down the lidar sensor cost by over 90% last year and that is without taking mass production savings into account. Thats about $7,000 before mass production.

Google’s Waymo invests in LIDAR technology, cuts costs by 90 percent

Velodyne most popular lidar today which almost every SDC company uses is the VLP-16 lidar and its now priced at $4,000.

Velodyne cuts VLP-16 lidar price to $4k - SPAR 3D

One of the attributes of Tesla fans most people dislike is the disregard of actual facts and reality. For example, you repeat the same talking points over and over again, no matter how outdated it is.

 

[Reciprocity]

Who told you that SDC taxis will cost $150k when deployed in mass?

$70,000? You are still using the same statistics from 10 YEARS AGO!
Waymo already cut down the lidar sensor cost by over 90% last year and that is without taking mass production savings into account. Thats about $7,000 before mass production.

Google’s Waymo invests in LIDAR technology, cuts costs by 90 percent

Velodyne most popular lidar today which almost every SDC company uses is the VLP-16 lidar and its now priced at $4,000.

Velodyne cuts VLP-16 lidar price to $4k - SPAR 3D

One of the attributes of Tesla fans most people dislike is the disregard of actual facts and reality. For example, you repeat the same talking points over and over again, no matter how outdated it is.

Who cares what the actual cost is when its 10-100x to much. You can quote the cost of a single piece of hardware and not understand what it takes to get it into a car and the computing power required to use it or any of the other thousands of attributes that do not have a specific sticker price. What Tesla fans really dislike is Audi forum trolls who come here an act like they know more then Tesla engineers, when they dont. No idea why you feel the need to troll around this forum and act like you speaking truths and facts when in fact you dont know what you dont know.

The complexity only goes up as you decrease the costs and you still need to decrease the cost of the lidar and supporting system by at least an order of magnitude, but more likely two orders of magnitude.

I agree with Elon that Lidar is the crutch that will kill the Waymos of the world. I am not saying that they cannot build a very expensive autonomous taxi, only that its pointless when compared to the bigger picture.

 

[RubberToe]

Autonomous cars hitting California roads in April...

General Discussion: 2018 Investor Roundtable

 

[malcolm]

Curious "partnership" for Velodyne:

The Future of Hope: Velodyne LiDAR and MADD Partner to Advance Safety on the Roadways - Velodyne LiDAR 360 Blog

Together, Velodyne and MADD will share their collaborative power to bring about the end of drunk driving deaths...........Velodyne LiDAR will never be drunk, drugged, distracted, or even drowsy.

Which seems to imply that the choice of sensor alone, solves the problem.

And I suppose if you have pre-mapped and learned routes between designated passenger collection points in town and drop-off points in residential areas then it would work.

But then an old trolley bus could do this too.

BTW I think the Trolley Song needs an extra verse update; something about the Trolley Problem

 

[mspohr]

Curious "partnership" for Velodyne:

The Future of Hope: Velodyne LiDAR and MADD Partner to Advance Safety on the Roadways - Velodyne LiDAR 360 Blog



Which seems to imply that the choice of sensor alone, solves the problem.

And I suppose if you have pre-mapped and learned routes between designated passenger collection points in town and drop-off points in residential areas then it would work.

But then an old trolley bus could do this too.

BTW I think the Trolley Song needs an extra verse update; something about the Trolley Problem

Last month a drunk Tesla owner tried to have AP drive him home. He got on the Bay bridge and it shut down and stopped on the road due to lack of attention. Good try, though.

 

[cwerdna]

Waymo's 360-degree demo ride shows what self-driving cars 'see' pointed to Waymo reaches 5 million self-driven miles – Waymo – Medium.

The Waymo odometer rolled over to 5,000,000 in February: our self-driving cars have now covered five million miles on public roads.

That’s a major milestone. Yet what’s really exciting for us is how we’re picking up speed. Our 1st million miles took six years to complete. Our 4th million miles took six months, and this 5th million took just under three months. Today we’re driving as many miles in one day as the average American adult drives in a whole year.

 

[daktari]

Nice video, almost like the FSD-video 1,5 years ago. "the person in the driver seat is only there for legal reasons..." Well, really hope Tesla is not far behind, but not to optimistic.
Btw, was on a short trip to Phoenix, on a jog one of these Waymo Pacificas stopped for me while crossing the street. Driver seemed surprised, but I do not know whether the car or the driver controlled it... Anyways, I could live with a car looking like that if it really could drive my son to the karate all by itself!

 

[MXFLA]

I just saw a Cadillac ad for super cruise that had a quote, (Tesla eat your heart out). Aurora, Aptiv, and others seem to be close to bringing out new products. Audi says they will have L4 highway pilot ready in 18 months. I have AP 1 so I’m not expecting great things out of my system in the future but am beginning to doubt this whole AP2 coast to coast driving thing. I watched Space X launch the Falcon heavy and Elon is a billionaire and I’m certainly not, but I’m beginning to wonder if GM or others are going to begin passing us by.

 

[lunitiks]

I’m beginning to wonder if GM or others are going to begin passing us by

 

[Bladerskb]

Who cares what the actual cost is when its 10-100x to much.

Who cares about the decreasing cost of every piece of tech ever made, just drink more koolaid from elon that has been promising level 5 fsd is here since 2015.

You can quote the cost of a single piece of hardware and not understand what it takes to get it into a car and the computing power required to use it or any of the other thousands of attributes that do not have a specific sticker price.

We know all the equipment required for sdc, unlike you there are people who actually reason and research and analyze and don't rely on hype pieces from elon musk.

What Tesla fans really dislike is Audi forum trolls who come here an act like they know more then Tesla engineers, when they dont.

The consensus from almost every researcher and sdc company is that elon is an idiot and he is playing people like u like a fiddle.

The complexity only goes up as you decrease the costs and you still need to decrease the cost of the lidar and supporting system by at least an order of magnitude, but more likely two orders of magnitude.I agree with Elon that Lidar is the crutch that will kill the Waymos of the world. I am not saying that they cannot build a very expensive autonomous taxi, only that its pointless when compared to the bigger picture.

ignorance is bliss.

Let's just swallow the kool aid from a guy whose statement on sdc has been wrong every single time. whose cars are disabled by the slight splash of dirty water (rearview cameras), whose radar is disabled by the smallest sign of snow. whose cameras are blinded by the sunlight. sure. sure. bottoms up!

 

[S4WRXTTCS]

I keep coming back to this: we know that vision, alone, is sufficient for driving. Every consumer car to date has been driven using vision alone. There may be questions around whether people can figure out the right type of processing to do, or whether some camera suite has good enough resolution/contrast, etc, but we know it’s possible.

Vision only would work if the expectation of safety from an autonomous car wasn't orders of magnitudes greater than a human driver.

Hitting a kid because the cameras were blinded by the sun won't be acceptable
Crashing into cars in the fog won't be acceptable

Well meaning drivers get into accidents every day because they just couldn't see. Why would we give autonomous cars this handicap?

Even non-autonomous cars need have better sensing capability to help drivers out. One of these is cross-traffic rear radar because the backup camera by itself isn't good enough.