Waymo

[diplomat33]

Waymo raised another $2.5B:

"As we build, deploy, and commercialize the Waymo Driver, it’s gratifying to see so much interest in being part of our journey to transform mobility, making it safer, more accessible, and more sustainable along the way. So we’re proud to announce our latest investment round of $2.5 billion, with participation from Alphabet, Andreessen Horowitz, AutoNation, Canada Pension Plan Investment Board, Fidelity Management & Research Company, Magna International, Mubadala Investment Company, Perry Creek Capital, Silver Lake, funds and accounts advised by T. Rowe Price Associates, Inc., Temasek, and Tiger Global. We’ll use this latest round of investment to continue advancing the Waymo Driver, as well as continuing to grow our team.

With tens of millions of miles driven on public roads across 25 U.S. cities, and tens of billions of miles driven in simulation, our experience has shown us, and our investors, the massive opportunity ahead. We’re building and deploying the Waymo Driver to serve riders, deliver parcels, move freight, and eventually, to empower personal car ownership.

We’re the first company to operate a fully autonomous, public ride-hailing service - Waymo One, and we’ve served thousands of rider-only trips in Metro Phoenix. We’ve been driving in San Francisco and the Bay Area since we first started in 2009, and most recently San Francisco residents are seeing the Waymo Driver on our growing fleet of all-electric Jaguar I-PACE vehicles throughout the city. We’re building the most advanced technology stack for urban driving, and we’re partnering with key community organizations as we shape our product."

Read more here:

Transforming mobility with the confidence of world-class investors

As we build, deploy, and commercialize the Waymo Driver, it’s gratifying to see so much interest in being part of our journey to transform mobility, making it safer, more accessible, and more sustainable along the way. So we’re proud to announce our latest investment round of $2.5 billion, with...

blog.waymo.com

 

[mark95476]

Nope. White vehicles and completely blank/unmarked. Autonomous vehicle markings are generally prominent so these were unusual and stood out. I'll try to grab a photo and/or save the dashcam footage next time.

All the I-PACEs I've seen in Arizona say "Waymo Engineering" on the side in a somewhat out of the way place, is there at least that?

 

[diplomat33]

Waymo greatly improved the resolution of the lidar from the 4th Gen hardware to the 5th Gen hardware.

Here is a lidar point cloud on the 4th Gen (like on the Pacificas in Chandler):

Here is a lidar point cloud on the 5th Gen (like on the I-Pace):

That is insane!

I got this from the recent presentation that Aguelov did at the AV Workshop at the University of Toronto. It's academic but he presents a lot of info on Waymo's most recent machine learning work:

Waymo presentation starts around the 3:34:00 mark:

 

[willow_hiller]

Here is a lidar point cloud on the 5th Gen (like on the I-Pace):

The thing I find most fascinating about these new images is that as lidar resolution increases, the output approaches visual camera output, but with distances instead of colors.

Definite advantages over vision only, including being able to see that full picture in the dark and accurately measuring distances without needing to train a neural network. Will definitely still need good headlights and high resolution cameras to read the text off of signs/flat road markings, though.

Do we have any information on how much the new hardware costs? Is it still the situation with the i-Paces that a large portion of the trunk space is occupied with computers?

 

[diplomat33]

The thing I find most fascinating about these new images is that as lidar resolution increases, the output approaches visual camera output, but with distances instead of colors.

Definite advantages over vision only, including being able to see that full picture in the dark and accurately measuring distances without needing to train a neural network. Will definitely still need good headlights and high resolution cameras to read the text off of signs/flat road markings, though.

Definitely. I think we see now why Waymo uses lidar.

But Waymo also uses high resolution cameras for the reasons you state. Here is what the blog on the 5th Gen hardware says about Waymo's cameras:

"With high-dynamic range and thermal stability over automotive temperature ranges, our vision system cameras are designed to capture more detail and provide sharper images in the toughest driving environments.

Our latest long range cameras and 360 vision system now see much farther than before, allowing us to identify important details like pedestrians and stop signs greater than 500 meters away. Through advanced design innovations, including custom lenses and precise optomechanical engineering, our vision systems enable much higher performance levels than cameras on cars today."

Source: Waypoint - The official Waymo blog: Introducing the 5th-generation Waymo Driver: Informed by experience, designed for scale, engineered to tackle more environments

Do we have any information on how much the new hardware costs?

We don't have exact numbers but here is what we know so far:

1) 5th Gen hardware is half the cost of the 4th Gen hardware:

Waymo blog: "our latest sensors deliver more performance than ever before, at half the cost of our previous generation."

2) Total cost (car + hardware) is no more than a "moderately equipped Mercedes S-Class.

Krafcik in an interview earlier this year:

"Let me paraphrase it like this: If we equip a Chrysler Pacifica Van or a Jaguar I-Pace with our sensors and computers, it costs no more than a moderately equipped Mercedes S-Class. So for the entire package, including the car - today."

Source: Waypoint - The official Waymo blog: Introducing the 5th-generation Waymo Driver: Informed by experience, designed for scale, engineered to tackle more environments

Source: Waymo CEO Says Tesla Is Not A Competitor, Gives Estimated Cost Of Autonomous Vehicles

Is it still the situation with the i-Paces that a large portion of the trunk space is occupied with computers?

I don't know. I could not find any specific info.

 

[Daniel in SD]

The thing I find most fascinating about these new images is that as lidar resolution increases, the output approaches visual camera output, but with distances instead of colors.

That image looks like they're showing reflectivity not distance.

Cruise claims the Origin will cost "roughly half the cost of what a conventional electric SUV costs today.” That sounds much cheaper.
If only there were a currency that they could use to quantify the cost. haha.

 

[JJRicks]

The thing I find most fascinating about these new images is that as lidar resolution increases, the output approaches visual camera output, but with distances instead of colors.

Definite advantages over vision only, including being able to see that full picture in the dark and accurately measuring distances without needing to train a neural network. Will definitely still need good headlights and high resolution cameras to read the text off of signs/flat road markings, though.

Do we have any information on how much the new hardware costs? Is it still the situation with the i-Paces that a large portion of the trunk space is occupied with computers?

In this post, they show the intention is to have the trunk open for storage. Waypoint - The official Waymo blog: Designing the 5th-generation Waymo Driver Would like to test this when the cars are introduced for passenger service, which... could be pretty soon?

https://twitter.com/x/status/1405024625931677697

 

[diplomat33]

Drone view of the Waymo depot in SF.

It seems Waymo has a bunch of robotaxis in SF ready to go. We can see some Pacificas but also a bunch of I-Pace's. Hopefully, we will see Waymo launch a commercial ride-hailing service in SF soon.

I am curious what the colored robotaxis are.

 

[diplomat33]

In this post, they show the intention is to have the trunk open for storage. Waypoint - The official Waymo blog: Designing the 5th-generation Waymo Driver Would like to test this when the cars are introduced to passenger service

Is it still the situation with the i-Paces that a large portion of the trunk space is occupied with computers?

From this pic, it seems the computer will only take up a small part of the trunk. The I-Pace will have a good size trunk space:

 

[JJRicks]

From this pic, it seems the computer will only take up a small part of the trunk. The I-Pace will have a good size trunk space:

That means there will be zero chance of me accidentally seeing the computer , like has happened previously (someone left the cover off)

 

[Dan D.]

Drone view of the Waymo depot in SF.



I am curious what the colored robotaxis are.

I would guess a Pride event in San Francisco in rainbow colours

 

[Doggydogworld]

Is it still the situation with the i-Paces that a large portion of the trunk space is occupied with computers?

It's easy to hardwire these algorithms in ASICs, reducing space and power requirements to trivial levels. It's just not worth the upfront cost until you are deploying 100s of thousands.

Drone view of the Waymo depot in SF.

The intrepid videographer actually stopped his car on the shoulder of 280 and got out to bring this to us! Here's the depot last November, when it first started operating and the ratio of Pacificas to i-Paces was reversed.

In other news, Waymo hit a lawsuit-seeking scooter rider in San Francisco last night. Waymo released a statement saying the safety driver had put it into manual mode (blame the driver, who says the whole world can't copy Tesla, ha). I wonder if they'll tell us what their simulation shows? The whole "we ran these accidents through our simulation and avoided every one" schtick is misleading, though. But it does seem Waymo's wrecks are mostly caused by humans. A safety driver accidentally disengaged while sleeping (!!) a few years ago and ended up in the median. Another disengaged when the van went to the left side of the lane unexpectedly, then got clipped by the lane-splitting motorcyclist the van was moving over to make space for. I wonder if they've reached the point where safety drivers cause more harm than they prevent?

 

[Daniel in SD]

It's easy to hardwire these algorithms in ASICs, reducing space and power requirements to trivial levels. It's just not worth the upfront cost until you are deploying 100s of thousands.


The intrepid videographer actually stopped his car on the shoulder of 280 and got out to bring this to us! Here's the depot last November, when it first started operating and the ratio of Pacificas to i-Paces was reversed.

In other news, Waymo hit a lawsuit-seeking scooter rider in San Francisco last night. Waymo released a statement saying the safety driver had put it into manual mode (blame the driver, who says the whole world can't copy Tesla, ha). I wonder if they'll tell us what their simulation shows? The whole "we ran these accidents through our simulation and avoided every one" schtick is misleading, though. But it does seem Waymo's wrecks are mostly caused by humans. A safety driver accidentally disengaged while sleeping (!!) a few years ago and ended up in the median. Another disengaged when the van went to the left side of the lane unexpectedly, then got clipped by the lane-splitting motorcyclist the van was moving over to make space for. I wonder if they've reached the point where safety drivers cause more harm than they prevent?

They seem to be saying that the safety driver was in manual mode well before the collision occurred.
Their safety report shows 29 simulated collisions and 18 actual collisions so their safety drivers are definitely still improving safety.

 

[diplomat33]

In other news, Waymo hit a lawsuit-seeking scooter rider in San Francisco last night. Waymo released a statement saying the safety driver had put it into manual mode (blame the driver, who says the whole world can't copy Tesla, ha). I wonder if they'll tell us what their simulation shows? The whole "we ran these accidents through our simulation and avoided every one" schtick is misleading, though. But it does seem Waymo's wrecks are mostly caused by humans. A safety driver accidentally disengaged while sleeping (!!) a few years ago and ended up in the median. Another disengaged when the van went to the left side of the lane unexpectedly, then got clipped by the lane-splitting motorcyclist the van was moving over to make space for. I wonder if they've reached the point where safety drivers cause more harm than they prevent?

The scooter incident was clearly the fault of the safety driver. He was driving the car manually well before the accident when it should have been in autonomous mode. It is very likely that the Waymo Driver would not have caused the accident. Waymo has shown before that the Waymo Driver can safely handle similar situations in simulations. And if the car had been in autonomous mode with an attentive safety driver, there is a very high chance, the accident would have been prevented. This seems like exactly the type of human error that Waymo is trying to reduce by developing safe autonomous driving.

I do think that Waymo should not allow safety drivers to drive manually unless there is a really good reason for it. Driving manually defeats the whole purpose of testing autonomous driving.

Eventually, when the Waymo Driver is proven safe enough in the SF area, Waymo should remove the safety drivers entirely, as they have done in limited ways in Chandler.

 

[rxlawdude]

The scooter incident was clearly the fault of the safety driver. He was driving the car manually well before the accident when it should have been in autonomous mode.

Funny. Any Tesla AP/FSD incident is clearly the fault of the DRIVER, too. So, Waymo is better why?????

 

[diplomat33]

Funny. Any Tesla AP/FSD incident is clearly the fault of the DRIVER, too. So, Waymo is better why?????

They are different situations:

Tesla is a driver assist so the driver is always responsible.

Waymo is better because it is autonomous. It does not need a driver when the autonomous mode is turned on. The driver is only responsible when the autonomous mode is turned off.

In this instance, the Waymo autonomous mode was turned off. The safety driver was driving the Waymo manually. Since the Waymo autonomous mode was off, the safety driver was responsible for the accident. But that is only because the autonomous mode was off.

 

[willow_hiller]

In this instance, the Waymo autonomous mode was turned off. The safety driver was driving the Waymo manually. Since the Waymo autonomous mode was off, the safety driver was responsible for the accident. But that is only because the autonomous mode was off.

There definitely is a difference in media coverage.

The only headline I can find is: Waymo driver hit a scooter in San Francisco, no injuries reported

Which makes it very clear the Waymo was not in autonomous mode. When Waymo says it was in manual mode, the media doesn't question them about that and doesn't leave any shadow of doubt in the reporting.

Meanwhile the media coverage of Tesla is much less forgiving. For e.g. Was Tesla Model S in Texas Crash on 'Autopilot'? Musk Denies

Say what you want about Tesla's credibility when it comes to deadlines, but I don't think Elon has done anything to merit the media openly questioning stated facts about a crash.

 

[Daniel in SD]

There definitely is a difference in media coverage.

To be fair if someone had died or been seriously injured the coverage would probably be different.
The driver of the Waymo vehicle is presumably still alive so Waymo would have a tough time falsely attributing the accident to them.
The coverage of the Texas incident was more of an issue of the media blindly trusting police reports.

 

[ThomasD]

Have they expanded their geofence areas with the new 5th generation hardware?

 

[diplomat33]

Have they expanded their geofence areas with the new 5th generation hardware?

Not yet. They are still testing the 5th Gen hardware. They have not deployed the 5th Gen on public rides yet.