Waymo Says Its Self-Driving Tech Is More Advanced Than Tesla FSD, Others

[TMC Staff]

Waymo, formerly Google’s autonomous driving project, doesn’t agree with Elon Musk’s comments on how advanced Tesla’s self-driving tech is. Waymo was launched back in 2009 as a Google-funded project to tackle the problem of creating self-driving vehicles. Now it’s owned by Google’s parent company, Alphabet, still chipping away at the autonomous vehicle problem and company...
[WPURI="https://teslamotorsclub.com/blog/2020/11/13/waymo-says-more-advanced-tesla-fsd-autonomous/"]READ FULL ARTICLE[/WPURI]

 

[diplomat33]

Waymo, formerly Google’s autonomous driving project, doesn’t agree with Elon Musk’s comments on how advanced Tesla’s self-driving tech is. Waymo was launched back in 2009 as a Google-funded project to tackle the problem of creating self-driving vehicles. Now it’s owned by Google’s parent company, Alphabet, still chipping away at the autonomous vehicle problem and company...
[WPURI="https://teslamotorsclub.com/blog/2020/11/13/waymo-says-more-advanced-tesla-fsd-autonomous/"]READ FULL ARTICLE[/WPURI]

Waymo is right. Waymo is years ahead of the competition. Waymo has the most advanced FSD hardware and software. Waymo also has deployed L4 autonomous cars, completely driverless, with no safety driver. No one else has done that.

Musk is also wrong about Waymo when he says that their FSD only works in a premapped 50 sq. mi area. Waymo's FSD is capable of working in many areas. Waymo has autonomous cars in 25 metro areas across the US. The 50 sq. mi area is just where Waymo has launched a commercial service. He is either uninformed about how Waymo's FSD works or is choosing to mischaracterize Waymo's FSD in order to put Tesla in a better light.

 

[XDriver007]

Waymo has it covered. Hire an army to keep maps updated every minute and pray it never rains or snows. BOOM, GOOGLE WINS!

 

[MikeHolliday]

Waymo is right. Waymo is years ahead of the competition. Waymo has the most advanced FSD hardware and software. Waymo also has deployed L4 autonomous cars, completely driverless, with no safety driver. No one else has done that.

Musk is also wrong about Waymo when he says that their FSD only works in a premapped 50 sq. mi area. Waymo's FSD is capable of working in many areas. Waymo has autonomous cars in 25 metro areas across the US. The 50 sq. mi area is just where Waymo has launched a commercial service. He is either uninformed about how Waymo's FSD works or is choosing to mischaracterize Waymo's FSD in order to put Tesla in a better light.

Waymo may have its benefits, but when it comes to being able to drive from Point A to Point B anywhere in the US I definitely believe that Tesla has the advantage...

 

[diplomat33]

Waymo may have its benefits, but when it comes to being able to drive from Point A to Point B anywhere in the US I definitely believe that Tesla has the advantage...

Tesla's main advantage is being able to mass produce cars and deploy them quickly. So if Tesla's software can achieve reliable autonomous driving, then Tesla will have a lot of autonomous cars on the road. But the software needs to get there.

 

[GoSharks]

At this point, Waymo is completely correct. They have driverless cars on public streets. Tesla is currently nowhere close to that.

 

[MichaelP90DL]

Yeah, good luck with that.

 

[Ralanham]

So just like gm super cruze the more you limit the higher the rating you can get.

 

[S4WRXTTCS]

Tesla's main advantage is being able to mass produce cars and deploy them quickly. So if Tesla's software can achieve reliable autonomous driving, then Tesla will have a lot of autonomous cars on the road. But the software needs to get there.

Don't forget the hardware.

 

[jeremymc7]

Doesn't matter if it's more advanced if you can't put it in a production car for under $200k, and without making bulky let alone killing the aerodynamics.

 

[S4WRXTTCS]

Waymo may have its benefits, but when it comes to being able to drive from Point A to Point B anywhere in the US I definitely believe that Tesla has the advantage...

Waymo can drive autonomous from way more point A to Point B's than Tesla.

Tesla can current drive autonomously ZERO miles.

It really comes down to either being L4 or not.

 

[S4WRXTTCS]

So just like gm super cruze the more you limit the higher the rating you can get.

Stairs have you know stairs
Hiking trails have switchbacks.
bikes for kids have training wheels.

Sometimes you limit things to make life easier.

The problem with Supercruise is its stuck. As an example Supercruise one can't get updated with a SW update to Enhanced Supercruise. There also isn't any indication that Enhanced Supercruise will ever morph into something L3 capable.

Now that doesn't mean Tesla will ever achieve L3 or better either. We simply don't know.

If Tesla doesn't achieve L3 or better than I'd have to say Supercruise is superior because it allows hands free L2 driving. Sure its limited to controlled access freeways, but that's the only place I use Autopilot.

 

[MikeHolliday]

I travel on a lot of 2 lane roads in Ohio and on these roads I am using Auto Pilot 90%+ of the time which I am driving. Am I alert, absolutely, but I am constantly amazed at the job that it does...

 

[diplomat33]

So just like gm super cruze the more you limit the higher the rating you can get.

GM's Supercruise is still L2 like Tesla. It does not get a higher rating because it is limited to certain highways. Waymo is L4 because the cars are capable of fully driving without a safety driver in a geofenced area.

 

[S4WRXTTCS]

GM's Supercruise is still L2 like Tesla. It does not get a higher rating because it is limited to certain highways. Waymo is L4 because the cars are capable of fully driving without a safety driver in a geofenced area.

I think they were referring to the rating by consumer reports.

But, most of that has to do with driver monitoring and proper hands free benefits.

Anyways L2 != L4 so its entirely irrelevant.

 

[diplomat33]

I think they were referring to the rating by consumer reports.

But, most of that has to do with driver monitoring and proper hands free benefits.

Anyways L2 != L4 so its entirely irrelevant.

Ok. I thought by rating, they meant it gets a higher SAE level.

And this thread is about Waymo saying they have better FSD than Tesla. Not sure what the Consumer Report rating has to do with that. CR did not give Waymo a rating.

But you are correct. CR actually rated Tesla's AP better in capability but knocked it down because of poor driver monitoring. Supercruise gets high marks for its solid driver monitoring that allows for hands-free driving.

 

[abasile]

Waymo may have its benefits, but when it comes to being able to drive from Point A to Point B anywhere in the US I definitely believe that Tesla has the advantage...

This is something I've wondered about. The "Waymo Driver" does best with HD maps and LiDAR, but like Tesla, Waymo uses camera "vision" (they call it "360 vision"), radar, and neural nets. The question is, how well would Waymo perform if they were limited to using only vision and radar? We know that Waymo can operate, at least to some degree, without HD mapping because they can tolerate cases where the maps don't quite match reality or where there are road obstructions such as construction activity.

It seems possible that Waymo's vision and radar systems are at least on par with Tesla's. I don't feel comfortable assuming that Waymo lags in vision/radar just because they normally use HD maps and LiDAR. Waymo has been working on the problem longer, presumably with access to all manner of resources within Alphabet/Google.

On the other hand, it's indisputable that Tesla has a major data collection advantage over Waymo, which if leveraged as well as Tesla appears to be doing, could put Tesla ahead of Waymo in reaching L5 autonomy. This is where Waymo's high vehicle costs are a huge disadvantage, as each Waymo robotaxi has to be subsidized and there's no way for them to reach Tesla's scale with their current approach. When it comes to solving the march of nines and dealing with corner cases, I don't see how simulations could be nearly as useful as real-world data.

If Waymo has good vision/radar capabilities, though, then it would seem that they could sort of follow Tesla's strategy by partnering with OEMs and having much cheaper sensors, like those in Tesla cars, installed in millions of new cars as they are manufactured. However, that's not as easy as it sounds, as data from those sensors would have to be stored and uploaded to Waymo. For the cars to operate autonomously, Waymo and the OEMs would have to carefully negotiate and implement standards, and over-the-air updates would be a necessity. This could turn into a lengthy process!

 

[cypho]

Apples to Oranges

I totally believe that Google/Waymo could sell a car that is an order of magnitude better than Tesla's autopilot. But they don't. You can buy a car with Tesla's autopilot today. You cannot buy a car with Waymo self-driving system (or install it on another car).

If Waymo wants the self-driving championship ring, they have to actually put it out on the market.

 

[Todd Burch]

This is pretty simple.

Is Waymo’s current system better than Tesla’s FSD in the areas where Waymo can operate? Of course. Nobody could refute that with a straight face. FSD requires a lot of intervention, Waymo’s system very little.

However, who will be the first to widespread publicly-available self-driving cars that can operate in any driving domain? Tesla.

Who will be the first to have full generalized autonomy? Tesla.

Why?

1. Only Tesla has the hardware packaged efficiently from an aerodynamic perspective and at a reasonable hardware cost for the masses. That hardware will continue to get more powerful with less power draw per flop.

2. Only Tesla has the data gathering capability to solicit enough data to train their system to handle true edge cases. (Billions of miles, millions of vehicles). You need that much driving to catch the truly rare edge cases. Plus the ability to actively and automatically solicit/seek edge cases from the fleet with rapid turnaround of that data back to the mothership for training.

3. Competitors must deal with limitations of LIDAR in rain/snow. Must solve vision anyway to read signs, distinguish between two non-static things that have similar LIDAR returns, etc.

4.Relying heavily on mapping means the system breaks down when the real world doesn’t match the map anymore. That’s high maintenance, bro.

5. Only Tesla can gather a statistically significant dataset to validate system safety with short turnaround.

So is Tesla currently in the lead? No. It’s a simple fact that Waymo disengagements are far less frequent than Tesla engagements, in Waymo’s operating area.

But Tesla will advance way faster and cover the edge cases way faster and validate safety with billions of miles way faster.

As a wise man once said, it’s the PACE of innovation that matters.

 

[diplomat33]

4.Relying heavily on mapping means the system breaks down when the real world doesn’t match the map anymore. That’s high maintenance, bro.

This point has been debunked. It has been established that Waymo cars do not break down if the map is wrong. Waymo cars can still drive if the maps are wrong. Waymo themselves have confirmed this.