Introducing GAIA-2 🌎Generative world modeling just stepped up a gear. GAIA-2 is the latest development of Wayve’s video-generative world model tailored for driving. GAIA-2 offers richer, more realistic, and highly controllable synthetic driving scenarios, accelerating Wayve’s path to safe driver assistance and automated driving at scale. Learn more about... GAIA-2 in our Blog: #GAIA2 #GAIA #EmbodiedAIshow more

5-camera view 🎦GAIA-2 is breaking new ground with this surround camera view that is reflective of a real-world camera array. Most importantly, the videos are temporally and spatially coherent, which is critical for safety tests. #GAIA2

Increased controllability 🎮 GAIA-2 offers more control over all scene features like the ego-vehicle behavior, the behavior of other road agents, and environmental factors like road structure, weather, and time of day. See how we can manipulate the time of day and weather in the video below. #GAIA2

Expanded Diversity! 🗺️ GAIA-2 is trained on a global dataset, multiple vehicle platforms, and various sensor configurations. That means GAIA-2 can generate driving scenes from Germany, UK, US, cars, vans, etc. #GAIA2

Generating Safety-Critical Scenarios! 🦺You might not need emergency braking everyday, but it is a safety-critical manoeuvre. Using GAIA-2, we can simulate multiple common safety-critical scenarios, generating rare and high-risk events that can be impractical or unsafe to capture through real-world data collection alone. #GAIA2

Learn how GAIA-2 accelerates Wayve’s path to safe driver assistance and autonomous driving at scale. Blog 🔗

Announcing: Our most advanced speech-to-text model goes beyond accuracy to capture the real-world complexity of human conversation and deliver reliable, source-of-truth audio data. Explore Universal-2 updates 👇

Sorry to keep asking this, radar? lidar? Or camera-only for future driverless operations?

After going to GTC, seeing the latest autonomous driving simulator from @waabi, and watching @Wayve 's progress, it's clear that Elon was right years ago: that every auto maker will eventually be able to do AI driving. That said, even if every one of them added that capability today (they aren't, most are years away from including the computers and cameras in car to do this) they don't have the fleet size to do a Robotaxi network. And traditional automakers aren't selling enough electric cars (which is where these things will be first included) to get to a fleet of millions of cars anytime soon. And the fleet size is very important for finishing off the job, and getting rid of mistakes that can't be predicted in simulators. The fleet size is also important for another range of things, including AI assistants. Right now there are dozens of Teslas on the Golden Gate Bridge, for instance. So, if a crash happens there, who will be able to tell you first about it? Even better, who will be able to route their Robotaxis around such a thing, or give more accurate predictions for how long it'll take you to get across the bridge this morning? Or, who will be able to show you near real time videos of what's actually happening on the bridge? Right now it's not obvious about why all that is so important, but by the end of next year it will be. Robots are about to radically change our world. The one who has the most robots on the street will have information advantages for years until the others are able to sell millions of robots for the street. When I was young, my best friend was killed in a car crash when he fell asleep. I've dreamed of a day when that no longer was going to happen, and that day is coming. But when Uber was invented right in front of me in a Paris snowstorm, it set in motion a new dream: of transportation as a service because now we can see that we can make transportation much better and cheaper. Which is what the @waymo customers keep telling me: they don't want to own another car or take a ride in a vehicle with another human in it (when I'm in San Francisco and see people get out of Waymos I interview them about their experience). If you come to San Francisco you see just how deeply Waymo is changing that city. The rides per month keep going up in an exponential way. Today that hasn't hit my neighborhood, or yours, but over the next three years it probably will (Uber went from an idea in a Paris snowstorm to something that people around the world used in three years). In five, we will be in a totally different transportation world and there will be winners and losers. But, really, the market for all that that matters is China. It has nine cities larger than New York in population. So whoever wins in China will have a huge operational advantage. Last week Tesla sold 17,000 vehicles there: all robots. And I've been watching, and talking to people on the street in China that are watching this all go down and Tesla's FSD is out performing the Chinese manufacturers, none of which have announced a vehicle without a steering wheel yet. Understanding all this makes me even more bullish on Tesla, since its Cybercab will be the most efficient (translates into cheapest) to operate per mile and have the best experience while doing so. This is why Tesla's stock hasn't crashed, and has gone up about 25% in the past week, even though liberals are angry at Elon Musk (that problem doesn't exist on the streets of China). It is the only non Chinese manufacturer that is doing well in China, which really is the only market that matters. Anyone see a real Robotaxi competitor coming that will be able to get 10s of millions of vehicles onto the world's streets by 2027? I don't. But am watching, and ready to change my thesis. That all said: the news from Wayve is hugely exciting because it will lead to fewer deaths on our world's roads and is amazing AI engineering. Congrats @Jamie_Shotton and team!

cool

Fascinating! A world model that learns to drive like a panda learns bamboo groves – through endless curious exploration of possible paths.

Not so many yet talking about @wayve_ai!