FSD V GM CRUISE

[Daniel in SD]

so you’re saying FSD is L5? WTF?!? FSD appears to only be striving for L3

How would an L3 robotaxi work? Ask the drunk guy in the backseat to climb into the front and take control?
It's clear that FSD is L5, it handles the entire driving task and can drive anywhere, it just isn't as safe as a human yet so it requires a safety driver.

ok cool. By the way, I have ocean beach front property for sell in Nevada for sell - pending massive earthquake that will split Cali from mainland.

I'll sell you a ticket to another galaxy pending my invention of faster than light travel!

 

[S4WRXTTCS]

FSD appears to only be striving for L3

FSD to me appears to striving for make believe L4.

If Tesla was striving for L3 they would have spent all the time improving upon freeway NoA, and not on City Streets.

 

[BooMan]

FSD to me appears to striving for make believe L4.

If Tesla was striving for L3 they would have spent all the time improving upon freeway NoA, and not on City Streets.

Not sure I follow. L3 or L4 doesn’t necessarily speak to hwy or city street scenarios rather differing responsibility levels for driver and OEM. Am I miss understanding your point?

 

[Daniel in SD]

Not sure I follow. L3 or L4 doesn’t necessarily speak to hwy or city street scenarios rather differing responsibility levels for driver and OEM. Am I miss understanding your point?

Most people think L3 highway is easier than city. For example Mercedes is releasing an L3 highway traffic jam system in Germany and hopefully California soon.
Though maybe an L3 city system is possible. All the current robotaxis use remote assistance which is sort of like L3 (except they all claim the remote assistance doesn’t drive the car, it just helps the planner deal with edge cases).

 

[Dewg]

After reading about Waymo and Cruise on these forums, I decided to take a look at some math. The crash numbers are from the last year and from NHTSA reports (sources at the end)

Waymo has a fleet of about 700 cars. According to the report they had 62 crashes. Their incident percentage is 8.86%

Cruise has a fleet of about 30 cars. They had 23 crashes in the report. Their incident percentage is 76.,66%. This is more understandable as they are just starting their testing.

Tesla has a fleet of about 2 million cars. Let's be conservative and say that only 10% of Tesla owners use AP, NoA, or FSD Beta. That's 200,000 cars. They had 273 crashes in the report. Their incident percentage is 0.14%. If we lower the number of cars to just FSD Beta participants, that's 100,000 cars. Their incident percentage is then 0.27%.

I think the sheer number of Teslas on the road is one of the reasons we see more YouTube videos of their accidents on AP/FSD Beta. They, Telsa, also do not spend a dime on advertising, which means media companies prioritize articles/video of anything going wrong in a Tesla. Google and GM spend lots of money on advertising, so it's a little harder to show negative press against them - don't bite the hand that feeds you.

I was surprised at the number of incidents reported by NHTSA on Waymo, as I found it hard to find any crash videos on YouTube for them - just a few from years ago. I wonder why we don't see many crash videos for Waymo or Cruise.

Sources:

NHTSA ADS Report: https://www.nhtsa.gov/sites/nhtsa.gov/files/2022-06/ADS-SGO-Report-June-2022.pdf

NHTSA ADAS Report: https://www.nhtsa.gov/sites/nhtsa.gov/files/2022-06/ADAS-L2-SGO-Report-June-2022.pdf

 

[Daniel in SD]

I was surprised at the number of incidents reported by NHTSA on Waymo, as I found it hard to find any crash videos on YouTube for them - just a few from years ago. I wonder why we don't see many crash videos for Waymo or Cruise.

Read the descriptions of all the Cruise and Waylon crashes here and I think you’ll see why.

Autonomous Vehicle Collision Reports - California DMV

Manufacturers who are testing autonomous vehicles need to report any collision that resulted in property damage, bodily injury, or death within 10 days of the incident.

www.dmv.ca.gov

 

[AlanSubie4Life]

Waylon

Is Elon buying Waymo too, and renaming it?

He is so amazing.

 

[Dewg]

Read the descriptions of all the Cruise and Waylon crashes here and I think you’ll see why.

Autonomous Vehicle Collision Reports - California DMV

Manufacturers who are testing autonomous vehicles need to report any collision that resulted in property damage, bodily injury, or death within 10 days of the incident.

www.dmv.ca.gov

Now this DMV site is making wonder why everyone wants to crash into Cruise and Waymo cars.

 

[Daniel in SD]

Now this DMV site is making wonder why everyone wants to crash into Cruise and Waymo cars.

Waymo claims to be driving 100,000 miles a week in San Francisco. Try to do that and dodge all the bad human drivers.
I do suspect that they drive in non human ways and get hit more often too.

 

[DJT1]

Waymo claims to be driving 100,000 miles a week in San Francisco. Try to do that and dodge all the bad human drivers.
I do suspect that they drive in non human ways and get hit more often too.

That's 10 miles/second around the clock for the fleet. Just how many Waymo's are there driving around SF... You would think they would be all over the place, but they are not.

Unless they mean "San Francisco Bay Area" and not just the city.

 

[AlanSubie4Life]

That's 10 miles/second around the clock for the fleet. Just how many Waymo's are there driving around SF... You would think they would be all over the place, but they are not.

Unless they mean "San Francisco Bay Area" and not just the city.

10 miles per minute. There are about 10k minutes in a week.

If you assume average speed of 5mph it’s about 120 vehicles I think, each doing 120 miles per day (note I am not saying anything about the number of vehicles).

 

[Daniel in SD]

That's 10 miles/second around the clock for the fleet. Just how many Waymo's are there driving around SF... You would think they would be all over the place, but they are not.

Unless they mean "San Francisco Bay Area" and not just the city.

https://twitter.com/x/status/1426339194767507459

Not sure. They reported 2.3 million miles of on road testing in California which is only 44k miles a week average. However it looks like they did the vast majority of testing towards the end of the year. They did it with 567 vehicles. It's weird to me that even the most used vehicle was 17k miles. Why not run them 24/7?

https://www.dmv.ca.gov/portal/file/2021-autonomous-mileage-reports-csv/

 

[AlanSubie4Life]

Why not run them 24/7?

Lidar is very unreliable.

 

[Daniel in SD]

Lidar is very unreliable.

They're using the 110V charging cable that came with the car.

 

[enemji]

So what happens when you need an extra 200,000 or more cars at once? Cat 4 Hurricane headed toward Louisiana and Mississippi. You need to evacuate 3 million people.

That is what buses are for. Cars take up unnecessary space for the number of passengers it can accommodate on a given road especially if it is an evacuation path.

 

[S4WRXTTCS]

Not sure I follow. L3 or L4 doesn’t necessarily speak to hwy or city street scenarios rather differing responsibility levels for driver and OEM. Am I miss understanding your point?

The point was where a person would start from to accomplish a given task.

L3 on controlled access highway is far easier than L3 driving on city streets. There is so much one has to account for when dealing with city streets, and things happen quickly.

There is far greater demand for Highways than City streets because the majority of the boring time that people want to get back is on highways.

 

[BooMan]

Most people think L3 highway is easier than city. For example Mercedes is releasing an L3 highway traffic jam system in Germany and hopefully California soon.
Though maybe an L3 city system is possible. All the current robotaxis use remote assistance which is sort of like L3 (except they all claim the remote assistance doesn’t drive the car, it just helps the planner deal with edge cases).

Ok? I guess ?? Just so long as people can distinguish fact (SAE Level definitions) vs random marketing (utterly useless non definitions that seem to shift). I guess I only understand levels in context of what they are (ie what the SAE says) and not what OEMs “say”

 

[DJT1]

Oops, another reason FSD Beta is different. Car fleet doesn‘t suddenly go inop.

Driverless Robotaxi Fleet Paralyzed for Hours in San Francisco

Apparently, the first system breakdown of a driverless robot taxi fleet occurred. In San Francisco, at least a dozen autonomous Chevrolet Bolts from GM Cruise Automation were spotted blocking the i…

thelastdriverlicenseholder.com

 

[S4WRXTTCS]

Oops, another reason FSD Beta is different. Car fleet doesn‘t suddenly go inop.

Driverless Robotaxi Fleet Paralyzed for Hours in San Francisco

Apparently, the first system breakdown of a driverless robot taxi fleet occurred. In San Francisco, at least a dozen autonomous Chevrolet Bolts from GM Cruise Automation were spotted blocking the i…

thelastdriverlicenseholder.com

Sadly this kind of thing doesn't bode well for any autonomous vehicle.

People don't realize how hard it is to make things free of bugs or issues. In fact I reckon that things of this nature will be almost a daily occurrence with widespread autonomous vehicle adoption.

When's the last time anyone used any complicated electronic device that didn't occasionally fail.

 

[DJT1]

Sadly this kind of thing doesn't bode well for any autonomous vehicle.

People don't realize how hard it is to make things free of bugs or issues. In fact I reckon that things of this nature will be almost a daily occurrence with widespread autonomous vehicle adoption.

When's the last time anyone used any complicated electronic device that didn't occasionally fail.

Individual failures are one thing, but fleet wide failures is a design problem.