Autonomous Car Progress

[diplomat33]

When you call for a Waymo, Cruise or Argo does it go through a dispatch center then they send the car that is closest to you? Or does the request go through a car and the car knows exactly where you are and knows where the car needs to go with zero human intervention.

It works like the Lyft or Uber app. You enter your intended destination and pay via the app. The app knows your current location. The software automatically tells the closest available car to pick you up. You get in, hit the "start ride" button and the car takes you to the destination you put into the app.

Will it be possible to have the car recognize voice commands and take you to your destination based on where you tell it to go and adjust price accordingly. Say I'm going to the Supermarket but once I am in the car I decide I want to go to the Walmart Supercenter. Will I be able to say take me to Walmart in this town?

No. I do not believe that Waymo currently responds to voice commands. You can change your destination but you need to do it from the app. Voice commands could be a future feature some day.

 

[Jeff N]

A sometimes amusing hit piece against self-driving AI.

Even After $100 Billion, Self-Driving Cars Are Going Nowhere

They were supposed to be the future. But prominent detractors—including Anthony Levandowski, who pioneered the industry—are getting louder as the losses get bigger.

www.bloomberg.com

 

[AlanSubie4Life]

A sometimes amusing hit piece against self-driving AI.

Even After $100 Billion, Self-Driving Cars Are Going Nowhere

They were supposed to be the future. But prominent detractors—including Anthony Levandowski, who pioneered the industry—are getting louder as the losses get bigger.

www.bloomberg.com

Yep, I agree, I think everyone is years away. I used to think Cruise and Waymo were close, maybe a year or so, but I really don’t think they are. Several years at least. If ever! We might not have a solution! It’s one thing to not run into or kill anyone (which may be hard). It’s quite another to have a workable service.

Many many years to go.

 

[Kablooie]

Yep, I agree, I think everyone is years away. I used to think Cruise and Waymo were close, maybe a year or so, but I really don’t think they are. Several years at least. If ever! We might not have a solution! It’s one thing to not run into or kill anyone (which may be hard). It’s quite another to have a workable service.

Many many years to go.

In my opinion Tesla is 90% there. Granted the last 10% will be difficult but they are making consistent improvements and haven’t even fully incorporated all the technology they plan to use yet. Teslas goal is much more ambitious than Cruise or Waymo because they want a consumer car that can drive anywhere, not just in premapped limited areas.

 

[AlanSubie4Life]

In my opinion Tesla is 90% there. Granted the last 10% will be difficult but they are making consistent improvements and haven’t even fully incorporated all the technology they plan to use yet. Teslas goal is much more ambitious than Cruise or Waymo because they want a consumer car that can drive anywhere, not just in premapped limited areas.

They are nowhere near that close. There are very few places that FSD can drive right now without monitoring.

I predict 2027 or so. And not sure what system that will be. Might be L2 still.

Remember they still are perfecting AP - it is not quite there yet (specifically it can’t handle traffic, branching/merging, construction, high curvature, cannot adopt arbitrary context-appropriate lane positioning, and cannot pick context-appropriate speeds ).

 

[diplomat33]

A sometimes amusing hit piece against self-driving AI.

Even After $100 Billion, Self-Driving Cars Are Going Nowhere

They were supposed to be the future. But prominent detractors—including Anthony Levandowski, who pioneered the industry—are getting louder as the losses get bigger.

www.bloomberg.com

Just the title seems a bit clickbait to me. AV companies like Waymo have deployed fully autonomous cars in several areas and adding new areas every year. Not to mention that we are much closer to solving full autonomy everywhere now than we were say 10 years ago. Not saying we are close to L5 but we are certainly closer than we were 10 years ago. So no, I would not say that we are "going nowhere".

I think part of the problem is that the public was given false expectations on fully autonomous cars, from movies but also from certain CEOs who misjudged how hard the problem would be. So people assumed we would have driverless everywhere by now (which was never realistic) and are now disappointed that *all* we have are robotaxis in some geofenced areas. But I like to see things as a glass half full. Robotaxis used to be scifi and now we actually have some real driverless robotaxis in limited areas!! If you understand how hard autonomous driving is, that is a huge accomplishment by itself. We have made a lot of progress in autonomous driving already and more to come.

Another thing: yes, robotaxis sometimes mess up, like we see with Cruise incidents. I think that was another false expectation that robotaxis were going to be these perfect human-like drivers right out of the gate. At this stage of development, robotaxis will not be perfect drivers. The tech will make mistakes, especially in the beginning. But the tech will get better.

 

[Jjrss]

Veoneer (Tier 1 ADAS supplier) just announced a partnership with Arbe Robotics, maker of the "Phoenix" high res imaging/perception 4D radar and 360 surround radars..

Looks like the partnership will be a direct competitor to Tesla auto pilot feautures.

Looking at Veoneer's current customers:

Mercedes-Benz
Nissan
Fiat/Chrysler/Dodge
GM
Ford
Volvo
Subaru
... list goes on.

Should also be mentioned the NHTSA Senior Advisor, and Tesla AP/FSD critic "Missy" Cummings was a board member of Veoneer up until Oct 2021 when she left the board to take the Presidential appointment to the NHTSA.

 

[eli_]

In my opinion Tesla is 90% there. Granted the last 10% will be difficult but they are making consistent improvements and haven’t even fully incorporated all the technology they plan to use yet. Teslas goal is much more ambitious than Cruise or Waymo because they want a consumer car that can drive anywhere, not just in premapped limited areas.

I don't think it's a fair comparison because they have totally different business models and products.

If Tesla was actually trying to make a Level 4 or 5 robotaxi today of course they'd have similar limits too. Real systems have real limits. If you want to sell investors on robotaxis with the current state of technology, you need to put the car inside a box and pre-map everything in 3D, otherwise it would break all the time like FSD does and it wouldn't be a very convincing demo. Investors may come away thinking that robotaxis are actually decades away instead of months and put their money elsewhere.

Tesla doesn't care about any of that, it's a completely different business model for them. All they gotta do is have some crazy features that work some of the time, and people will pay $15k for it, and it'll help sell Teslas too. I showed the FSD beta to some coworkers and they all came away thinking that Tesla is so advanced and ahead of the rest of the car industry, they were totally enamored with the 3D visualizations on there. "Look it shows the bus on there, wow!"

 

[Soda Popinski]

I don't think it's a fair comparison because they have totally different business models and products.

If Tesla was actually trying to make a Level 4 or 5 robotaxi today of course they'd have similar limits too. Real systems have real limits. If you want to sell investors on robotaxis with the current state of technology, you need to put the car inside a box and pre-map everything in 3D, otherwise it would break all the time like FSD does and it wouldn't be a very convincing demo. Investors may come away thinking that robotaxis are actually decades away instead of months and put their money elsewhere.

Tesla doesn't care about any of that, it's a completely different business model for them. All they gotta do is have some crazy features that work some of the time, and people will pay $15k for it, and it'll help sell Teslas too. I showed the FSD beta to some coworkers and they all came away thinking that Tesla is so advanced and ahead of the rest of the car industry, they were totally enamored with the 3D visualizations on there. "Look it shows the bus on there, wow!"

It is an interesting take on Tesla's "business model" - in that they do state a goal of robo-taxies, but unlike every robo-taxi company, are already getting revenue for the unfinished product. If they do get to robo-taxies, then that's even more revenue.

My uninformed prediction is Tesla won't achieve Level 5 without at least one more hardware revision and without Dojo up and running. There's a possibility a new "FSD" chip prototype was running in Optimus, so that aspect may be sooner than people think.

 

[diplomat33]

Cruise AV shows superhuman reaction time and avoids a collision:

https://twitter.com/x/status/1579601826638622720

 

[Daniel in SD]

Cruise AV shows superhuman reaction time and avoids a collision:

https://twitter.com/x/status/1579601826638622720

I have to wonder if the human driver only did this because he thought Cruise would brake in time. Impressive though.
I hope Elon fixes Twitter’s video quality…

 

[diplomat33]

I have to wonder if the human driver only did this because he thought Cruise would brake in time.

Nah, I would not give the human driver that much credit. More likely, the human driver is just a bad driver and got lucky that the other car was an AV that could brake in time.

 

[Daniel in SD]

Nah, I would not give the human driver that much credit. More likely, the human driver is just a bad driver and got lucky that the other car was an AV that could brake in time.

I see a lot of collisions in the reports that seem to be drivers attempting to pass AVs so it wouldn’t surprise me if people turn left in front of them very aggressively. I wonder if AVs will have to start reporting other drivers for doing illegal dangerous maneuvers.

Btw none of the recent reported Cruise “collisions” including the one where the cyclist was yelling at the vehicle are in the collision reports.

 

[diplomat33]

I see a lot of collisions in the reports that seem to be drivers attempting to pass AVs so it wouldn’t surprise me if people turn left in front of them very aggressively. I wonder if AVs will have to start reporting other drivers for doing illegal dangerous maneuvers.

Yeah, I definitely think that some human drivers are disrespectful to AVs. I am just not sure they are thinking about the reaction time of the AV.

 

[ThomasD]

It is good that it missed the other car. I had this happen to me over the weekend except it happened at highway speed. I got real lucky I was going under the speed limit by 5. I wonder if the Cruise was traveling at 55 mph and someone coming in the opposite turned left in front of it if it would have missed the other car

 

[Doggydogworld]

I have to wonder if the human driver only did this because he thought Cruise would brake in time.

It's a staggered intersection. There's a prominent crosswalk Cruise had to cross before entering the intersection. The oncoming driver probably thought Cruise had a stop sign and the crosswalk paint included a stop line.

If the oncoming driver was trying to bully Cruise he wouldn't have suddenly stopped when he realize Cruise had proceeded into the intersection. Also, it's one thing to be aggressive around an AV, but you have to be pretty stupid to blatantly break the law right in front of a car festooned with cameras,

 

[Daniel in SD]

It's a staggered intersection. There's a prominent crosswalk Cruise had to cross before entering the intersection. The oncoming driver probably thought Cruise had a stop sign and the crosswalk paint included a stop line.

If the oncoming driver was trying to bully Cruise he wouldn't have suddenly stopped when he realize Cruise had proceeded into the intersection. Also, it's one thing to be aggressive around an AV, but you have to be pretty stupid to blatantly break the law right in front of a car festooned with cameras,

I swear half the AV collision reports state that the other drive fled the scene! This is even when there is a safety driver. People don’t seem concerned with breaking the law on video in San Francisco.

One thing to note is a human driver might have swerved left which would have allowed them to avoid collision even with a slower reaction time.

 

[AlanSubie4Life]

I swear half the AV collision reports state that the other drive fled the scene! This is even when there is a safety driver. People don’t seem concerned with breaking the law on video in San Francisco.

One thing to note is a human driver might have swerved left which would have allowed them to avoid collision even with a slower reaction time.

Yes, seems like a human could have avoided it, but would definitely have to be very alert and on their game.

This is definitely superhuman. I have yet to see anything like this from Tesla FSD, but I am sure it is coming soon.

 

[stopcrazypp]

It's a staggered intersection. There's a prominent crosswalk Cruise had to cross before entering the intersection. The oncoming driver probably thought Cruise had a stop sign and the crosswalk paint included a stop line.

If the oncoming driver was trying to bully Cruise he wouldn't have suddenly stopped when he realize Cruise had proceeded into the intersection. Also, it's one thing to be aggressive around an AV, but you have to be pretty stupid to blatantly break the law right in front of a car festooned with cameras,

As mentioned, judging by how people drive around Teslas they know are loaded with cameras (there are channels dedicated to showing this on YouTube), I highly doubt many care. I should also note there is zero indication Cruise is providing video footage to law enforcement even when accidents do occur (for example the report for the recent left turn accident appears to be all done on scene reconstruction and witness accounts).

 

[stopcrazypp]

Yes, seems like a human could have avoided it, but would definitely have to be very alert and on their game.

This is definitely superhuman. I have yet to see anything like this from Tesla FSD, but I am sure it is coming soon.

I had to dig a bit to find it again, but there was a video years ago of AP doing the same thing in similar conditions. Tesla was travelling 45mph at night and someone turned left in front of the car, and car brakes to save the day (driver claimed he did not touch brakes at all).

Tesla's autopilot helps Seattle Uber driver avoid a car crash

Seattle Uber driver Jon Hall was saved from crashing into an oncoming car thanks to Tesla's new autopilot feature

www.cbsnews.com

There are plenty of videos if you look up compilation videos of Teslas saving the day, with car stopping just in time to avoid crashing into a car that is rapidly crossing in front (usually red light runners) although those are typically in daytime. They also look "superhuman".