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.
Uh, that quote is not saying Waymo has "solved FSD". "Solved the fundamentals of FSD" is not the same as "solved FSD". Fundamentals refer to the basic, core ideas of something. I am saying that Waymo has solved the basic parts of FSD. But Waymo has not solved the more complex parts of FSD yet. So, Waymo has not solved all of FSD yet.
Ok.

Without figuring out how to scale, you haven’t figured out the “fundamentals” - IMO. Just like dozens of battery startups that solved lithium-air etc “fundamentally”. Or for that matter - Tesla has fundamentally solved FSD, they have to just reduce error rates now.

But that’s a different conversation.
 
  • Like
Reactions: mangrove79
So you'd agree with Elons "feature complete" phrasing on FSD I guess :)
To me, the basic step in FSD should be crash avoidance.

Tesla achieved crash avoidance technology by recruiting attentive drivers.

Waymo achieved crash avoidance technology by using technology and not staffing human drivers in its current robotaxis in Chandler, AZ.

That's the basic step: The crash avoidance technology must be good enough to remove human drivers from the car.

Crash avoidance technology is achievable with current science but the next step is very difficult: Intelligence.

An example of needing intelligence is when Waymo encountered an out-of-place traffic cone that's inconsistent with its pre-mapping, it did not collide with the cone: That's the easy part, but the next difficult part is intelligence: How to deal with a new scenario and it didn't know what to do next. Human technicians had to be dispatched to retrieve the car.
 
  • Like
Reactions: daktari
To me, the basic step in FSD should be crash avoidance.

Tesla achieved crash avoidance technology by recruiting attentive drivers.

Waymo achieved crash avoidance technology by using technology and not staffing human drivers in its current robotaxis in Chandler, AZ.

 
And about twice that if you count accidents that the safety drivers avoided. In Waymo's defense that was over 6.1 millions miles and included collisions as minor as a pedestrian walking in to the side of the car and 14 of the actual collisions were the Waymo vehicle getting rear ended.
 
Thanks for the link.

The article is very informative but it lumps all accidents whether Waymo hits others or others hit it and the confusing whether the collision did actually happen or just "simulated".

When I said collisions, I mean the Autonomous Vehicle is the one that is the collider, the culprit that moves to hit the victim. When the Autonomous Vehicle is a victim because others move and hit it then I don't count it as collisions at the automation's system fault.


An example is:

"Waymo cars were involved in one actual and two simulated events (i.e., events triggered by a disengagement) in which a pedestrian or cyclist struck stationary Waymo cars at low speeds."

In this scenario, Waymo was hit by a pedestrian or cyclist. The pedestrian or cyclist was not hit by Waymo.

An example is when a distracted pedestrian walked and smashed his face into the side of a moving firetruck with loud siren. To me, that's not the fault of the firetruck but this article would count that as a collision too!


1640288021787.png

Notice, in California, any interruption with the Autonomous system is called "disengagement". This article call those disengagement as "simulated events (i.e., events triggered by a disengagement)".

This article calls collisions "actual" and "simulated" which makes the readers think that there are more actual collisions.

An example when Waymo is not the culprit but as victim in rear-end collisions:

"Waymo reported 11 actual rear-end collisions involving its cars and one simulated collision. In eight of the actual collisions, another car struck a Waymo car while it was stopped; in two of the actual collisions, another car struck a Waymo car moving at slow speeds; and in one of the actual collisions, another car struck a Waymo car while it was decelerating. The simulated collision modeled a Waymo car striking a decelerating car."

Actual: 11 including:
8: Waymo was stationary, others hit it from the rear
2: Waymo was at slow speed and others hit it from the rear
1: Waymo was decelerating and others hit it from the rear

Simulated: 1
The model predicted that if Waymo kept decelerating at the current rate, the other would hit it from the rear. But it never occurred because it's a model, it's simulated.

In all, of the rear-end collisions example above (actual or simulated), Waymo is the victim and not the one who hits others.

So, first thing first, Autonomous Vehicles need to have the capability to avoid hitting others first.

Then the next step is intelligence: How to make sure others don't hit them.

Like in the case of the firetruck, first it should not hit others. But others like the distracted pedestrian was walking toward the firetruck, maybe it should spray water at the pedestrian to wake him up.
 
Last edited:

The last major obstacle to cracking the code on self-driving cars​


Instead, MobilEye wants to map the world’s roads as crib sheets for self-driving cars. “The idea behind the high-resolution maps is that you prepare all this information in advance,” Shashua explained. That way autonomous vehicles can already know there’s a “no right on red” sign at the next intersection, rather than relying on a computer vision system to spot the sign, read it, and apply its meaning to the road on the fly. “Once you connect to an actionable high-definition map, the performance accuracy is an order of magnitude better than if you don’t have a map,” argues Shashua. That makes a big difference in building regulators’ and the public’s trust in self-driving cars and eventually securing regulatory approval and winning customers.

 
  • Like
Reactions: diplomat33

The last major obstacle to cracking the code on self-driving cars​


Instead, MobilEye wants to map the world’s roads as crib sheets for self-driving cars. “The idea behind the high-resolution maps is that you prepare all this information in advance,” Shashua explained. That way autonomous vehicles can already know there’s a “no right on red” sign at the next intersection, rather than relying on a computer vision system to spot the sign, read it, and apply its meaning to the road on the fly.
LOL. Static worldview in a dynamic world.

ps : In my place, 90% of the cars turn on red when there is a "no turn on red". I'm sure if Tesla implemented this, there would be clamoring for an option to override it ;)
 
  • Like
Reactions: superblast


Big Tech’s Next Monopoly Game: Building the Car of the Future​

ps: see link of MapBox complaints
I prefer the use of HD Maps from TomTom which will be used in the next EV models of VW, Stellantis (PSA&FCA), Toyota/Lexus, Hyundai/KIA, etc and some Chinese OEM’s like NIO, Xpeng, BYD, Aito, etc.
BYD and Aito are using HarmonyOS from Huawei and Huawei is using TomTom Maps in their Petal Maps as well in their phones, watchers and EV cars. After Trump decided that Huawei was not allowed Google anymore.

I think the best for Tesla is when Apple would-be buy it, start a new brand, platform and approach with HD Maps.

So Elon can take a rest and try to find a new challenge ……after he refunded al those Tesla owners who paid $10k in advance for the FSD. 😁
 
You’ll notice there is a new sheriff in town cleaning up the posts here. Give me some time to get ‘er done.

Feel free to send me a private message with concerns and/or to re-report old posts that need cleanup.

For context, I own FSD on HW3 but have so few expectations about it that I haven’t bothered upgrading from MCU1.

Like all moderators, I have little tolerance for name calling. Respect, facts and punctuation go a long ways.
 
Last edited:
And you know that how ? Atleast have some intellectual honesty.

I said IMO. So, I am giving my opinion that I think their approach is the right one. From everything, I've seen from Mobileye, I feel like they are tackling the problem of FSD in a smart way. They have excellent camera vision, with redundant lidar and radar, that gives the car accurate perception of both the static and dynamic world. They have crowdsourced HD maps which will boost reliability. And they have good driving policy, called RSS, that has codified rules to help ensure safer driving. So they have a good solution for all the necessary pieces of FSD: perception, prediction, planning, and driving policy with a mind towards safety. But I was very honest that it was just my opinion. So how am I not being intellectually honest?

Anyway - not sure why every thread is turned into HD Map / LiDAR thread by Waymo fans. There are enough threads about those already - post there. Here let us talk about FSD 10.8.

I agree that threads should stay on topic. It is not my intention to turn others discussion into HD maps or LIDAR. I simply responded to Elon's BS claim about sensor fusion. Others bring up HD maps or LIDAR and I will respond, especially if they have a question that I think I can answer.

I think it is the "six degrees of Kevin Bacon" effect. Invariably, HD maps and LIDAR will come up somehow since they are so important to autonomous driving. Almost all AV companies use HD maps and LIDAR as a critical part of how their autonomous works. So it is kind of hard to talk about autonomous driving without talking about them. And Waymo is a leader in the AV field, doing serious work on autonomous driving. It is kind of hard to talk about autonomous driving without talking about companies like Waymo who are pioneers in the field. That would be like talking about smart phones but not being allowed to talk about Apple. We discuss the work Tesla is doing in autonomous driving. We should also discuss the work that others companies are doing too.
 
Last edited:
Big news! Waymo is partnering with Geely to add the Waymo Driver to one of their vehicles:


I find this interesting because we know that Geely is adding the Mobileye SuperVision on their Zeekr sedan cars. So it looks like Geely is going with Mobileye for their sedan cars but going with Waymo for this minivan EV.

Also, in the video, we can see that the vehicle has no steering wheel or pedals. So it would appear that this vehicle will be a true driverless vehicle.
 
Last edited:
Big news! Waymo is partnering with Geely to add the Waymo Driver to one of their vehicles:


I find this interesting because we know that Geely is adding the Mobileye SuperVision on their Zeekr sedan cars. So it looks like Geely is going with Mobileye for their sedan cars but going with Waymo for this minivan EV.

Also, in the video, we can see that the vehicle has no steering wheel or pedals. So it would appear that this vehicle will be a true driverless vehicle.

I actually think that this is just a separate partnership and is not representative of Geely’s own autonomy plans.
 
  • Helpful
Reactions: diplomat33
I actually think that this is just a separate partnership and is not representative of Geely’s own autonomy plans.

That's possible. I am more excited for what this deal means for Waymo. The vehicle looks ideal for ride-hailing with lots of people and it has no steering wheel or pedals. So it would mean Waymo's first truly driverless robotaxi since the firefly pod years ago.
 
Big news! Waymo is partnering with Geely to add the Waymo Driver to one of their vehicles:


I find this interesting because we know that Geely is adding the Mobileye SuperVision on their Zeekr sedan cars. So it looks like Geely is going with Mobileye for their sedan cars but going with Waymo for this minivan EV.

Also, in the video, we can see that the vehicle has no steering wheel or pedals. So it would appear that this vehicle will be a true driverless vehicle.
That is not my take.

I think Geely will make some robotaxis with Waymo. Looks more like earlier Volvo partnership - though that came to nothing.
 
Big news! Waymo is partnering with Geely to add the Waymo Driver to one of their vehicles:


I find this interesting because we know that Geely is adding the Mobileye SuperVision on their Zeekr sedan cars. So it looks like Geely is going with Mobileye for their sedan cars but going with Waymo for this minivan EV.

Also, in the video, we can see that the vehicle has no steering wheel or pedals. So it would appear that this vehicle will be a true driverless vehicle.
Geely's having similar partnerships/investments everywhere. Not sure if it was pointed out in this thread, but Geely actually has a joint venture with Baidu called Jidu Auto (with Baidu having a controlling 55% stake) where they plan to make EVs. They showcased a car with the same idea (no steering wheel or pedals) back in August:
AI-packed 'robocar' puts Baidu in the self-driving fast lane

The Waymo venture looks more like it's another contract manufacturing deal similar to what they had for the Pacifica and iPace and not something necessarily that would be integrated into Geely as core tech they will use for their other vehicles.
 
The Waymo venture looks more like it's another contract manufacturing deal similar to what they had for the Pacifica and iPace and not something necessarily that would be integrated into Geely as core tech they will use for their other vehicles.

Correct. I am not suggesting that this is Geely's attempt at FSD. It is clear to me that Geely made a deal to produce a new robotaxi vehicle for Waymo. Like I wrote earlier, I like this news because it shows us that Waymo has a new robotaxi vehicle planned.
 
Correct. I am not suggesting that this is Geely's attempt at FSD. It is clear to me that Geely made a deal to produce a new robotaxi vehicle for Waymo. Like I wrote earlier, I like this news because it shows us that Waymo has a new robotaxi vehicle planned.
Has waymo addressed the question of whether they would ever be involved in self-driving cars that are designed for consumer purchase as opposed to robotaxi fleet usage?