Welcome to Tesla Motors Club
Discuss Tesla's Model S, Model 3, Model X, Model Y, Cybertruck, Roadster and More.
Register

Autonomous Car Progress

This site may earn commission on affiliate links.
I think 99% of the people here who have a beef with Elon are only upset because Tesla took their money for FSD years ago and will probably never deliver what was promised.
Those that have a beef with Tesla over the FSD package have no-one else to blame but themselves.

No matter what time or which "version" of FSD you were buying, the site had always indicated that this is a work in progress... here is an example of that verbiage on the site today:
1626100758203.png



But in reality the issue is exacerbated by folks like you -- who didn't even BUY the package, but go to extreme lengths telling others how they should feel about their purchase.
 
Last edited:
Unfortunately, when you posted the link, the forum software included that "joe's" comment. That's what I hate about the auto-embed feature here. It adds things that you didn't necessarily want to include (sometimes the goal is just to post a link).
Yes, but 1 - I want the source linked, 2 - the context of the post is clear beyond without confusion.
The problem is the dimplomat had nothing useful to add, so he latches on to anything that might make Waymo look bad and tries to steer the conversation off in the direction with plenty of Waymo marketing material to boot. (this was just the latest example)
 
Those that have a beef with Tesla over the FSD package have no-one else to blame but themselves.
Well, most people don't like blaming theirselves when they think they've been deceived, it only makes them more angry!
But in reality the issue is exacerbated by folks like you -- who didn't even BUY the package, but go to extreme lengths telling others how they should feel about their purchase.
I think most people's feelings on the matter are valid. The only position I don't understand is people who were ok with leasing FSD for three years and getting absolutely nothing, those people should definitely be upset!
verbiage on the site today
The site's verbiage has changed over time and we've had this argument a few times. I agree that the current verbiage really doesn't promise much at all. There's still the issue that the CEO of the company talks with create confidence about some product to be released in the near future that is also called FSD and people assume that that's what they're ordering without actually reading the fine print.
For people who want to read endless arguing about the three different revisions to the order page:
 

First International safety standard for fully automated driving systems has been published​


The standard focused on L4 Low Speed Autonomous Driving (LSAD):

In the ISO 22737 standard, the group have set out the specific minimum safety and performance requirements for LSAD systems, providing a common language to help facilitate the development and safe deployment of this technology worldwide. The group included experts from Japan, USA, Canada, Australia, South Korea, China, Germany, France, the Netherlands, Hungary and the UK.

Some of the key points from the new standard include:

· Providing minimum operating capabilities for LSAD systems including guidance on Operational Design Domain (ODD) definition

· Guidance on how LSAD systems may fit into the wider transport ecosystem

· Performance requirements for different aspects of the LSAD system such Dynamic Driving Task; Emergency maneuvers (e.g. emergency stop and minimal risk maneuvers); hazardous situation identification; static and dynamic obstacle detection and avoidance

· Test procedures for various system functionalities

“The new LSAD standard puts a line in the sand for all OEMs to exceed and so enables a global market where all players rise to the same exacting standards as each other, keeping the public safe and the industry competitive."

The standard itself is behind a paywall. If I am able to purchase, I can share more details. But it is good news that we have our first AV Safety Standard

 
  • Like
Reactions: Dan D.

First International safety standard for fully automated driving systems has been published​


The standard focused on L4 Low Speed Autonomous Driving (LSAD):

In the ISO 22737 standard, the group have set out the specific minimum safety and performance requirements for LSAD systems, providing a common language to help facilitate the development and safe deployment of this technology worldwide. The group included experts from Japan, USA, Canada, Australia, South Korea, China, Germany, France, the Netherlands, Hungary and the UK.

Some of the key points from the new standard include:

· Providing minimum operating capabilities for LSAD systems including guidance on Operational Design Domain (ODD) definition

· Guidance on how LSAD systems may fit into the wider transport ecosystem

· Performance requirements for different aspects of the LSAD system such Dynamic Driving Task; Emergency maneuvers (e.g. emergency stop and minimal risk maneuvers); hazardous situation identification; static and dynamic obstacle detection and avoidance

· Test procedures for various system functionalities

“The new LSAD standard puts a line in the sand for all OEMs to exceed and so enables a global market where all players rise to the same exacting standards as each other, keeping the public safe and the industry competitive."

The standard itself is behind a paywall. If I am able to purchase, I can share more details. But it is good news that we have our first AV Safety Standard

Low-speed, is this limited to 25 mph (~40 kph)?
 
The problem is the dimplomat had nothing useful to add, so he latches on to anything that might make Waymo look bad and tries to steer the conversation off in the direction with plenty of Waymo marketing material to boot. (this was just the latest example)

Wrong. I replied exactly to what you posted. I responded to the substance of what Karpathy said. But what you posted was wrong. So I tried to correct the misinformation.
 
  • Disagree
Reactions: mikes_fsd
Low-speed, is this limited to 25 mph (~40 kph)?

I don't know. But considering that the examples they give are the robotaxi "pods", the limit is probably around 25 mph (40 ph) yes. But I would have to read the standard itself to confirm the exact speed.

EDIT: The standard says 32 kph or less.

"Low speed: the speed of an LSAD system shall be equal to or less than 8,89 m/s or 32 km/h"
 

First International safety standard for fully automated driving systems has been published​


The standard focused on L4 Low Speed Autonomous Driving (LSAD):

In the ISO 22737 standard, the group have set out the specific minimum safety and performance requirements for LSAD systems, providing a common language to help facilitate the development and safe deployment of this technology worldwide. The group included experts from Japan, USA, Canada, Australia, South Korea, China, Germany, France, the Netherlands, Hungary and the UK.

Some of the key points from the new standard include:

· Providing minimum operating capabilities for LSAD systems including guidance on Operational Design Domain (ODD) definition

· Guidance on how LSAD systems may fit into the wider transport ecosystem

· Performance requirements for different aspects of the LSAD system such Dynamic Driving Task; Emergency maneuvers (e.g. emergency stop and minimal risk maneuvers); hazardous situation identification; static and dynamic obstacle detection and avoidance

· Test procedures for various system functionalities

“The new LSAD standard puts a line in the sand for all OEMs to exceed and so enables a global market where all players rise to the same exacting standards as each other, keeping the public safe and the industry competitive."

The standard itself is behind a paywall. If I am able to purchase, I can share more details. But it is good news that we have our first AV Safety Standard

For what it's worth, here is a preview copy of some of the standard. I wouldn't quote from it in this format.

 

Attachments

  • ISO-PRF-22737.pdf
    1.2 MB · Views: 44
  • Like
Reactions: diplomat33
For what it's worth, here is a preview copy of some of the standard. I wouldn't quote from it in this format.


here is the link to the pdf of the preview that is more readable.

 
Thanks. I checked the preview of the standard that you showed me and the standard says up to 32 KPH.
This standard allows L4 on predetermined routes only. Other standards of course cover different ODDs.

Seems Aurrigo in particular is hoping for L5 at some point too based on this photo.
Aurrigo-in-Canada.jpg



Their shuttle aims for 50kph and a wider ODD, but could still fall under LSAD in some uses.

These kind of vehicles could be used in the Vegas Convention Centre Loop Tunnel.
Auto-Shuttle_550x300px.jpg
 
Last edited:
  • Like
Reactions: diplomat33
This standard allows L4 on predetermined routes only. Other standards of course cover different ODDs.

That is correct. This standard is only for predetermined routes. But at least it is a start for standardizing different types of AVs.

I am still waiting for IEEE P2846 to be released since it will cover ODDs for all speeds, ride-hailing on public roads. So it will be more relevant to automakers like Tesla or AV companies like Waymo.
 
That is correct. This standard is only for predetermined routes. But at least it is a start for standardizing different types of AVs.

I am still waiting for IEEE P2846 to be released since it will cover ODDs for all speeds, ride-hailing on public roads. So it will be more relevant to automakers like Tesla or AV companies like Waymo.
I hope IEEE P2846 is inclusive of all driving parameters. At the moment they list an "example" speed of 55mph, but we know people will vastly exceed whatever sensible limits are established. I trust they will take into account willful excess of acceptable guidelines.
 
  • Like
Reactions: diplomat33
AutoX unveiled their Gen5 robotaxi at an event a few days ago:

Summary:

- 28 automotive grade camera sensors generating more than 200 million pixels per frame 360-degrees around the car.

- 6 high-resolution lidar sensors that produce 15 million points per second

- Surround 4D radar.

- 2 NVIDIA DRIVE Ampere architecture GPUs that deliver 900 TOPS each with max of 2200 TOPS combined.

"The Shenzhen fully driverless robotaxi service is just the first stop in AutoX’s roadmap to deploy a global driverless vehicle platform.

With a population of more than 12 million people and ranking in the top 50 of global cities with the heaviest traffic, Shenzhen provides an ideal setting for developing a scalable robotaxi model.

The startup plans to roll out thousands of autonomous vehicles powered by the Gen5 system over the next couple of years and expand to multiple cities around the world. AutoX is working with partners such as Stellantis and Honda to integrate their technology in a variety of vehicle platforms."

Source:
 
AutoX unveiled their Gen5 robotaxi at an event a few days ago:

Summary:

- 28 automotive grade camera sensors generating more than 200 million pixels per frame 360-degrees around the car.

- 6 high-resolution lidar sensors that produce 15 million points per second

- Surround 4D radar.

- 2 NVIDIA DRIVE Ampere architecture GPUs that deliver 900 TOPS each with max of 2200 TOPS combined.

"The Shenzhen fully driverless robotaxi service is just the first stop in AutoX’s roadmap to deploy a global driverless vehicle platform.

With a population of more than 12 million people and ranking in the top 50 of global cities with the heaviest traffic, Shenzhen provides an ideal setting for developing a scalable robotaxi model.

The startup plans to roll out thousands of autonomous vehicles powered by the Gen5 system over the next couple of years and expand to multiple cities around the world. AutoX is working with partners such as Stellantis and Honda to integrate their technology in a variety of vehicle platforms."

Source:
I imagine that doesn't sell for less than $50K USD, does it? SIX lidars? 28 cameras?

Can one say "overkill" without exaggeration?
 
No, they couldn't- because it takes years and years to get the mapping done.
This is a myth. When Waymo wants to test in a city they spend a couple weeks driving around to build maps and learn local customs and quirks. Google mapped the whole world pretty cheaply, and updates most areas every year or two (and yes, those cars have had lidar for many years). I've not come across anyone in the autonomous driving field who thinks maps are a bottleneck.

But hindsight is 20/20. It's easy to look back and say Waymo should have done things differently. But back in 2017, when Waymo started the service in Chandler, the FSD was not as good as it is today and the costs were a lot higher than they are now. So at the time, it probably did not look like they could expand yet.
They obviously did think they could expand it because they "ordered" 82,000 cars.

I'm not criticizing Waymo's initial approach. It's almost impossible to get the business model right on the first try. That's why entrepreneurs iterate like crazy. Waymo doesn't.

But Waymo still has the best FSD software and hardware. I say that because Waymo is a top innovator in ML. Waymo showed NN that is better than the current state of the art. Waymo has the best disengagement rate in CA. Waymo also custom builds all their sensors and they are among the best in the industry.
That's why I said "mostly". They've blown most of their tech lead, but not all of it. At least not yet.

And nobody in the US has duplicated the Waymo service in Chandler yet.
But they're doing it in China. And Cruise may beat Waymo in SF, a US market that actually matters.

But I think Waymo needs to make a big move very soon.
Ain't gonna happen.

Robotaxis lose money because they are so expensive. What robotaxis are good for is as a real-world development platform. I think that Waymo has not scaled up to more cities because a development platform does not need more cities.
Waymo didn't order 82,000 "development platforms" in early 2018. Initial upfront cost doesn't matter if your business model works. Payting 100k instead of 50k for your first 100,000 Robotaxis costs you an extra 5 billion, but deploying at scale gives you a 500b market cap.

That extra 50k spread over 500k miles is only a dime a mile, anyway. It doesn't affect first mover P&L. Furthermore, the cost would drop rapidly after the first few thousand units (cue the Wright's Law choir). So it only ends up costing you a couple billion extra. It's an absolute no brainer, IF you have customers. Waymo doesn't. And doesn't know how to get them.
 
  • Helpful
Reactions: diplomat33
I imagine that doesn't sell for less than $50K USD, does it? SIX lidars? 28 cameras?

It is NOT designed to be sold to the customer, it is designed to be a robotaxi that you summon on your phone and ride in to go somewhere! So, it does not need to be cheap enough for the public to buy. It just need to be cheap enough that they can make a profit with enough rides.

Can one say "overkill" without exaggeration?

It is not overkill IMO. When designing a safe robotaxi, it is better to have too much perception than not enough. Nobody will blame you for having too many sensors when you don't kill a pedestrian. They will blame you for not having enough sensors if your robotaxi fails to detect and accidentally kills a pedestrian.
 
It is NOT designed to be sold to the customer, it is designed to be a robotaxi that you summon on your phone and ride in to go somewhere! So, it does not need to be cheap enough for the public to buy. It just need to be cheap enough that they can make a profit with enough rides.



It is not overkill IMO. When designing a safe robotaxi, it is better to have too much perception than not enough. Nobody will blame you for having too many sensors when you don't kill a pedestrian. They will blame you for not having enough sensors if your robotaxi fails to detect and accidentally kills a pedestrian.
So you're touting technology that is (a) unproven and (b) likely unable to be cost justified. You also haven't explained how all those data will actually be processed onboard. We've discussed that having too much resolution complicates computational bandwidth.

I laugh at your last statement. Are you 100% sure that having this many sensors would preclude accidentally killing a pedestrian? I'd love some objective citation that guarantees 100% this touted technology won't kill a pedestrian. And when/if it does, indeed the robotaxi will be blamed.
 
So you're touting technology that is (a) unproven and (b) likely unable to be cost justified.

How is it unproven tech? Many AV companies (Waymo, AutoX, Baidu, Cruise, Pony.AI etc) use HD cameras, radar and lidar and the sensors are absolutely proven to work well.

In terms of cost, obviously, we don't know exactly what AutoX's costs are. But Waymo uses similar sensors and they have a cost per mile of only $0.30 per mile. So that gives us an idea. It is certainly not a given that they can't justify the costs.

I laugh at your last statement. Are you 100% sure that having this many sensors would preclude accidentally killing a pedestrian? I'd love some objective citation that guarantees 100% this touted technology won't kill a pedestrian. And when/if it does, indeed the robotaxi will be blamed.

I never said it would never kill a pedestrian. The point is to minimize the chance of the robotaxi hitting a pedestrian to as close to 0% as possible. Certainly, with 360 coverage with both high grade cameras, high grade radar and high grade lidar, I think the odds of the perception not detecting the pedestrian are going to pretty close to zero.
 
Last edited: