Waymo

[Doggydogworld]

I was referring to this line which implies the I-Pace also takes up storage for the computer.

"Both vehicles use an extensive array of cameras and sensors to collect information about their surroundings and the robust computer system that runs them takes up storage space in both vehicles."

Images from Waymo's Gen5 blog post show they still use some of the trunk space.

I agree that it is unlikely that Waymo is maintaining two separate code bases. But how do you explain that the 5th Gen drives so much better than the 4th gen? Is the massive improvement from 4th Gen to 5th Gen simply because of the newer sensors or is the improvement also from better software?

They improve continually, but what evidence do we have that Gen5 in SF drives 'so much better' than Gen4 in Chandler today? Recent SF videos curated by Waymo vs. year-old JJRicks full-truth videos is apples vs. oranges.

It is possible they froze the Chandler code, of course. They won't tell us one way or the other. I wonder what they plan to do with all those Pacificas, anyway?

 

[diplomat33]

Images from Waymo's Gen5 blog post show they still use some of the trunk space.

I am going by the quote from the Waymo Product Manager on the Road to Autonomy podcast. Here is the quote:

"The vehicles operating at Sky Harbor will be the Jaguar I-PACEs which have 25.3 cubic feet of cargo space. This space can hold 5 roller carry-on bags and will be available for travelers to store their bags on the journey to their destination, as the Waymo compute stack has gotten significantly smaller and more efficient over the years."

The quote indicates to me that travelers will be able to use the 25.3 cu. ft. of cargo space of the I-Pace.

Waymo's Rider-Only Experience | The Road to Autonomy

Sonca Nguyen discusses Waymo's rider-only experience.

www.brulteco.com

They improve continually, but what evidence do we have that Gen5 in SF drives 'so much better' than Gen4 in Chandler today? Recent SF videos curated by Waymo vs. year-old JJRicks full-truth videos is apples vs. oranges.

Well, we have videos of the I-pace handling driverless in more difficult driving environments than the 4th Gen Pacificas. So that suggests better driving. Plus, we know that the 5th Gen hardware is significantly better than the 4th Gen hardware. We also know that Waymo has made significant software improvements since the rides in the 4th Gen that JJ Ricks filmed.

It is possible they froze the Chandler code, of course. They won't tell us one way or the other. I wonder what they plan to do with all those Pacificas, anyway?

Yeah. I don't know what Waymo plans to do with the Pacificas. It is possible they will just keep them around for ride-hailing in Chandler while they expand with the 5th Gen everywhere else. I still think they will eventually discontinue them, especially when the Geely vans is deployed since it could fill the same role as the Pacificas.

 

[diplomat33]

Waymo employee mentions has taken 4 hour driverless rides in downtown Phoenix during rush hour with no incident.

https://twitter.com/x/status/1578459850287087617

 

[diplomat33]

Waymo is answering questions on Instagram. Here are a couple I thought were interesting:

That last sentence "we look forward to bringing capabilities from the trucking side to our ride-haling service" implies to me that Waymo will do longer trips with their robotaxis at some point. But it seems for now, they are focused on short trips in cities.

This seems like a typical non-answer. I can't tell if they are saying "we considered it and did some tests but it is not a good fit" or are they being coy? Personally, I do hope that Waymo deploys ride-hailing in NYC. That would be a very lucrative market for robotaxis.

 

[powertoold]

Is there a deep dive video or document that goes into Waymo's technology?

 

[diplomat33]

Is there a deep dive video or document that goes into Waymo's technology?

Yes. I don't think there is one single video. But Waymo has shared lots of info on their tech in different reports and tech presentations.

Here are 2 tech presentations that go pretty deep in the ML:

This is the official website for the Waymo Driver that provides some info on how the autonomous driving works.

Self-Driving Car Technology for a Reliable Ride - Waymo Driver

The Waymo Driver is our self-driving car technology. Get a safe, reliable ride from The World's Most Experienced Driver.

waymo.com

There are also some good blogs on the tech.

The Waymo Driver Handbook: Teaching an autonomous vehicle how to perceive and understand the world around it

Imagine you are driving through downtown San Francisco and a cyclist, traveling against the flow of traffic, cuts right in front of you. Or maybe you are navigating a narrow, two-way street with a car heading the opposite direction, and you nudge over for them to pass. Perhaps, you are driving...

blog.waymo.com

Utilizing key point and pose estimation for the task of autonomous driving

Historically, computer vision relies on rigid bounding boxes to locate and classify objects within a scene; however, one of the limiting factors in detection, tracking, and action recognition of vulnerable road users, such as pedestrians and cyclists, is the lack of precise human pose...

blog.waymo.com

Simulation City: Introducing Waymo's most advanced simulation system yet for autonomous driving

At Waymo, we use simulation to advance the Waymo Driver's abilities in several ways. We can amplify our number of real-world miles to accelerate our Driver’s learning, prepare for rare events, train new models, validate new software, and even evaluate how the Waymo Driver would have performed in...

blog.waymo.com

Informing smarter lidar solutions for the future

Lidar — which helps the Waymo Driver perceive the world around it in 3D — has become more common than ever before. Car manufacturers are integrating it into driver-assist systems; phone manufacturers are using it to enhance augmented reality capabilities, and robot vacuums even rely on it to...

blog.waymo.com

How we ensure the Waymo Driver operates safely in San Francisco

Last year, Waymo expanded its testing in San Francisco and began welcoming new riders in the city. As we continue to expand our operations providing rides to more San Franciscans, we’re sharing even more on how we evaluate the safety of our technology and operations in the City by the Bay.

blog.waymo.com

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

At Waymo, we’re building the World's Most Experienced Driver - our combination of hardware, software, and compute that powers vehicles to safely get people and things where they’re going.

blog.waymo.com

 

[diplomat33]

You can also go here and read a bunch of Waymo's latest ML papers on various subjects from perception, prediction to planning:

Autonomous Vehicle Research - Our Latest Publications - Waymo

Explore our autonomous vehicle research and Waymo Open Dataset. Our self-driving car research supports our innovative autonomous technology solutions.

waymo.com

 

[diplomat33]

Video of driverless ride in downtown Phoenix:

Also, here is an interesting little tidbit on how Waymo is relying more on ML for their planning:

A few months ago, Waymo rolled out an important improvement to its self-driving system. The specific change was the planning piece (as in planning and perception) of the software, according to senior product manager Pablo Abad. This newer version leans more heavily on machine learning and neural nets than it has in the past.

Adad explains: “Coding for all these different scenarios more manually can take time, as you can imagine,” Adad said. “As you build up more and more and more of these heuristics, it becomes harder for other developers to come in and tweak certain parameters without affecting other parts of the system. So instead, you give the system all of this training data, allow it to learn the best behavior in certain situations on its own, rather than having to manually get in there and insert heuristics.”

The result is a system that can handle more dynamic situations and improves more quickly. During my ride, the vehicle was able to smoothly execute trickier situations (like double parked vehicles) than I remember. It also allows Waymo to more easily fold new features in. For instance, as I was exiting, the robotaxi gave me a visual and audio alert that another vehicle was approaching from behind to ensure I didn’t whip my door open to wide on the busy street. It will also alert riders to upcoming cyclists.

Driverless in Phoenix and GM takes its EV offensive beyond vehicles | TechCrunch

The Station is a newsletter dedicated to all things transportation. We take a driverless ride in Phoenix and dig into GM's new energy unit.

techcrunch.com

 

[Doggydogworld]

Video of driverless ride in downtown Phoenix:

3 point turn at 4:40. At first I thought max speed was limited to 18 mph or something, but I eventually saw 30 and 35 on the broader boulevards.

 

[ThomasD]

Has Waymo expanded to include all communities in the San Francisco Metro area?

 

[diplomat33]

Has Waymo expanded to include all communities in the San Francisco Metro area?

I don't think so. AFAIK, Waymo operates in a select geofenced part of SF. But Waymo does operate 24/7.

 

[enemji]

The distinguishing technical questions are: "is a pre-computed mapping of the environment to ~cm level in 3d necessary for Waymo cars to function? Is a similar resolution direct physical measurement hardware at drive time, interacting with that mapping data, necessary to operate?"

I don't know what it is now, but once upon a time there had to be something like it because it came out of the 2007 DARPA challenge when there wasn't remotely either the computer science algorithmic knowledge or the computational hardware capability to do modern machine learning based vision processing.
I guess sort of similar to TERCOM used by older cruise missiles pre-GPS, correlations vs pre-mapped terrain.

Everything else is required tasks of any ADAS and semantic tags are different from cm level pre-mapping.

Now, Waymo might not need that now and surely does modern ML on vision, but could it operate successfully with low resolution maps of streets only (no buildings/overpasses/terrain mapped) ?

So Samir is the real-time driver making decisions WYSIWYG whereas the mapping is the navigator crying out. J/K

 

[diplomat33]

4 hr real-time, unedited, driverless ride in SF:

 

[ThomasD]

I was expecting a lot more heavy traffic in the videos.

 

[diplomat33]

 

[diplomat33]

Waymo just announced that LA will be their next ride-hailing area.

 

[Doggydogworld]

Why keep adding cities instead of scaling up in one city? Answer: they can't afford to scale up. So this is the only way they can show "progress" to investors.

In other news, Waymo recently gave a "driverless" ride to a San Francisco journalist. It didn't go great. First, it wasn't driverless. Maybe because of fog? The chaperone didn't really say. And Waymo says Gen5 can handle fog. There were also phantom braking events and the car got so confused at a drop-off point the safety driver had to put it into manual mode for a few blocks.

 

[Bitdepth]

Why keep adding cities instead of scaling up in one city? Answer: they can't afford to scale up. So this is the only way they can show "progress" to investors.

Or, hear me out.

Different cities provide different challenges not only in terms of autonomous driving but in running mobility as a service. Operating at small scale in different cities give them the opportunity to not only test and validate their software but also work out the issues and learn how to operate in different regions.

In other news, Waymo recently gave a "driverless" ride to a San Francisco journalist. It didn't go great. First, it wasn't driverless. Maybe because of fog? The chaperone didn't really say. And Waymo says Gen5 can handle fog. There were also phantom braking events and the car got so confused at a drop-off point the safety driver had to put it into manual mode for a few blocks.

There is a literal 4 hours long video of waymo driving around without a safety personnel above.

 

[diplomat33]

Why keep adding cities instead of scaling up in one city? Answer: they can't afford to scale up. So this is the only way they can show "progress" to investors.

No, that's just FUD. Waymo is scaling. That is why they are adding LA. I thought you all wanted Waymo to scale. They are doing it. But now when they do scale, you complain that they are scaling horizontally instead of vertically. The anti-Waymo crowd always have to find a negative slant. The fact that Waymo is adding LA is proof they are making real progress.

Waymo is doing exactly what they said they would do: pick cities that are a good match for the Waymo Driver, that add driving experience and that are good markets for ride-hailing.

Waymo CEO explains why they picked LA. They are building on the experience they have gained in SF and Phoenix. They believe LA will offer a geography well suited for the Waymo Driver, plus it is a good market for ride-hailing.

https://twitter.com/x/status/1582786645148938241

In other news, Waymo recently gave a "driverless" ride to a San Francisco journalist. It didn't go great. First, it wasn't driverless. Maybe because of fog? The chaperone didn't really say. And Waymo says Gen5 can handle fog. There were also phantom braking events and the car got so confused at a drop-off point the safety driver had to put it into manual mode for a few blocks.

Again, that is a very negative slant. If you read the blog, the ride was very good. There were only a couple issues. Obviously, AVs are not perfect. But Waymo is doing a lot of driverless rides that work great. And 5th Gen does work in fog.

 

[diplomat33]

Different cities provide different challenges not only in terms of autonomous driving but in running mobility as a service. Operating at small scale in different cities give them the opportunity to not only test and validate their software but also work out the issues and learn how to operate in different regions.

Exactly. Waymo has explained that they pick cities that can teach the Waymo Driver and also teach them about running a ride-hailing service. Right now, Waymo is focused on generalizing the Waymo Driver so they need to expose the Waymo Driver to different driving environments. They can always scale up in a city later.