Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Waymo World Model

waymo.com

Post reply on HN
Print view View original

371–380 of 699 posts

Re: The Waymo World Model

#371

Post by DiggyJohnson »

hungryhobbit wrote: post #325
hosh wrote: post #302

Earlier quoted context omitted.

I also think the presence of Sergey Brin has been making a difference in this.

Please, Google was terrible about using the tech the had long before Sundar, back when Brin was in charge. Google Reader is a simple example: Googl had by far the most popular RSS reader, and they just threw it away. A single intern could have kept the whole thing running, and Google has literal billions, but they couldn't see the value in it. I mean, it's not like being able to see what a good portion of America is…

How is this relevant? At best it’s tangentially related and low effort

Re: The Waymo World Model

#372

Post by inkysigma »

ses1984 wrote: post #368
yummypaint wrote: post #337

By leveraging Genie’s immense world knowledge, it can simulate exceedingly rare events—from a tornado to a casual encounter with an elephant—that are almost impossible to capture at scale in reality. The model’s architecture offers high controllability, allowing our engineers to modify simulations with simple language prompts, driving inputs, and scene layouts. Notably, the Waymo World Model generates high-fidelity,…

You could train it in simulation and then test it in reality.

Would it actually be a good idea to operate a car near an active tornado?

Re: The Waymo World Model

#373

Post by IMTDb »

yummypaint wrote: post #337

By leveraging Genie’s immense world knowledge, it can simulate exceedingly rare events—from a tornado to a casual encounter with an elephant—that are almost impossible to capture at scale in reality. The model’s architecture offers high controllability, allowing our engineers to modify simulations with simple language prompts, driving inputs, and scene layouts. Notably, the Waymo World Model generates high-fidelity,…

The idea is that, over time, the quality and accuracy of world-model outputs will improve. That, in turn, lets autonomous driving systems train on a large amount of “realistic enough” synthetic data.

For example, we know from experience that Waymo is currently good enough to drive in San Francisco. We don’t yet trust it in more complex environments like dense European cities or Southeast Asian “hell roads.” Running the stack against world models can give a big head start in understanding what works, and which situations are harder, without putting any humans in harm’s way.

We don’t need perfect accuracy from the world model to get real value. And, as usual, the more we use and validate these models, the more we can improve them; creating a virtuous cycle.

Re: The Waymo World Model

#374

Post by burgreblast »

hungryhobbit wrote: post #325
hosh wrote: post #302

Earlier quoted context omitted.

I also think the presence of Sergey Brin has been making a difference in this.

Please, Google was terrible about using the tech the had long before Sundar, back when Brin was in charge. Google Reader is a simple example: Googl had by far the most popular RSS reader, and they just threw it away. A single intern could have kept the whole thing running, and Google has literal billions, but they couldn't see the value in it. I mean, it's not like being able to see what a good portion of America is…

I never get the moaning about killing Reader. It was never about popularity or user experience.

Reader had to be killed because it [was seen as] a suboptimal ad monetization engine. Page views were superior.

Was Google going to support minimizing ads in any way?

Re: The Waymo World Model

#375

Post by smallmancontrov »

elgenie wrote: post #343
smallmancontrov wrote: post #251

Earlier quoted context omitted.

Humans: 70 deaths in 7 billion miles FSD: 2 deaths in 7 billion miles Looks like FSD saves lives by a margin so fat it can probably survive most statistical games.

Isn't there a great deal of gaming going on with the car disengaging FSD milliseconds before crashing? Voila, no "full" "self" driving accident; just another human failing [*]! [*] Failing to solve the impossible situation FSD dropped them into, that is.

Nope. NHTSA's criteria for reporting is active-within-30-seconds.

https://www.nhtsa.gov/laws-regulations/standing-general-orde...

If there's gamesmanship going on, I'd expect the antifan site linked below to have different numbers, but it agrees with the 2 deaths figure for FSD.

Re: The Waymo World Model

#376

Post by smallmancontrov »

Fricken wrote: post #364
hn_acc1 wrote: post #288

Earlier quoted context omitted.

Is that the official Tesla stat? I've heard of way more Tesla fatalities than that..

I don't know what he's on about. Here's a better list: https://en.wikipedia.org/wiki/List_of_Tesla_Autopilot_crashe...

Autopilot is the shitty lane assist. FSD is the SOTA neural net.

Your link agrees with me:

> two that NHTSA's Office of Defect Investigations determined as happening during the engagement of Full Self-Driving (FSD) after 2022.

Re: The Waymo World Model

#377

Post by tapoxi »

schiffern wrote: post #362
mattlondon wrote: post #89

Suddenly all this focus on world models by Deep mind starts to make sense. I've never really thought of Waymo as a robot in the same way as e.g. a Boston Dynamics humanoid, but of course it is a robot of sorts. Google/Alphabet are so vertically integrated for AI when you think about it. Compare what they're doing - their own power generation , their own silicon, their own data centers, search Gmail YouTube Gemini wor…

>I've never really thought of Waymo as a robot in the same way as e.g. a Boston Dynamics humanoid, but of course it is a robot of sorts. So for the record, with this realization you're 3+ years behind Tesla. https://www.youtube.com/watch?v=ODSJsviD_SU&t=3594s

Aren't they still using safety drivers or safety follow cars and in fewer cities? Seems Tesla is pretty far behind.

Re: The Waymo World Model

#378

Post by moffkalast »

ra7 wrote: post #22

The novel aspect here seems to be 3D LiDAR output from 2D video using post-training. As far as I'm aware, no other video world models can do this. IMO, access to DeepMind and Google infra is a hugely understated advantage Waymo has that no other competitor can replicate.

It's not unheard of, there are a handful [0] of metric monodepth methods that output data that's not unlike a really inaccurate 3D lidar, though theirs certainly looks SOTA.

[0] https://github.com/YvanYin/Metric3D

Re: The Waymo World Model

#379

Post by inkysigma »

parliament32 wrote: post #353
aaaalone wrote: post #341

Earlier quoted context omitted.

They probably just look at the results of the generation. I mean would I like a in-depth tour of this? Yes. But it's a marketing blog article, what do you expect?

> just look at the results of the generation And? The entire hallucination problem with text generators is "plausible sounding yet incorrect", so how does a human eyeballing it help at all?

I think because here there's no single correct answer that the model is allowed to be fuzzier. You still mix in real training data and maybe more physics based simulation of course but it does seem acceptable that you synthesize extremely tail evaluations since there isn't really a "better" way by definition and you can evaluate the end driving behavior after training.

You can also probably still use it for some kinds of evaluation as well since you can detect if two point clouds intersect presumably.

In much a similar way that LLMs are not perfect at translation but are widely used anyway for NMT.

Re: The Waymo World Model

#380

Post by cyanydeez »

robotresearcher wrote: post #250
cyanydeez wrote: post #245

Earlier quoted context omitted.

ok, but a care is a few meters wide, isn't that enough for driving depth perception similar to humans

The depths you are trying to estimate are to the other cars, people, turnings, obstacles, etc. Could be 100m away or more on the highway.

ok, but the point trying to be made is based on human's depth perception, but a car's basic limitation is the width of the vehicle, so there's missing information if you're trying to figure out if a car can use cameras to do what human eyes/brains do.
Post reply on HN