The AV industry is hitting a "Data Wall."
Brute force training on petabytes of data is hitting a dead end. The better the models get, the harder they are to improve.
At https://t.co/0KMFfBqkkU, we bypassed this wall with Factored Embodied AI—achieving Zero-Shot autonomous steering with just 1,000 hours of driving data.
Here is how we broke the barrier: 🧵
Big news: https://t.co/0KMFfBqkkU has entered a multi-year joint development agreement with @Honda 🎉
Together, we’ll accelerate Honda’s next-gen self-driving capabilities, including its Navigate on Autopilot (NOA) platform.
See thread for details 👇
#ADAS#AutonomousDriving #AI
Introducing GenSim-2: https://t.co/0KMFfBqkkU’s latest generative AI model for video editing in autonomous driving development!
GenSim-2 enables automakers to create and modify video data. The model supports dynamic weather and lighting adjustments (e.g., rain, fog, snow, glare, day/night), object customization (e.g., road surfaces, vehicles, pedestrians, buildings, vegetation), and ensures realism across multi-camera perspectives.
GenSim-2 enables the creation of fully AI-generated video scenes and augmented reality modifications of real video footage at up to 696x696 image resolution and up to 30 fps, supporting rare and complex corner case scenarios critical for testing and validation.
Trained using our proprietary Deep Teaching™ methodology, GenSim-2 generates highly realistic video data tailored to specific requirements. From enriching datasets to resolving rare corner cases, it’s a scalable, cost-effective solution for autonomous driving development.
Learn more in the the thread below ⬇️
Today, https://t.co/0KMFfBqkkU is featured in @nikkei Mobility!
The article highlights our approach to autonomous driving technology development, including in-vehicle software and simulation technology based on generative AI and Deep Teaching, and explains how our offering enables various automakers to more easily incorporate AI throughout the AV stack.
OEMs have recognized that software is the key differentiator for vehicle sales in the coming years and the path to high margins. However, they don't have large software-defined fleets already in market to support machine learning development, nor are they positioned to solely rely on fully end-to-end AI systems due to validation and interpretability challenges.
We enable OEMs to accelerate their roadmap and reduce costs by bridging the gap between simulation and real data with AI-powered virtual fleets, and we employ a hybrid approach that leverages modularity in our AI stack along with end to end technologies.
Full article (in Japanese): https://t.co/oWa8exRSGi
@therealarkin They have $$$ and a solid ML team. If this is to replace existing search functionality embedded into different meta properties -- it does not sound too crazy.
Israeli universities are offering spots to people in US universities who don’t feel safe. This is circulating at @MIT right now. People will certainly take them up on the offer.
There is a Jewish brain drain going on elite US universities. We should all be concerned about the future of technological innovation in this country when some of our greatest minds are being pushed off campus for being Jews.
Say what you want to say but X community notes is the best news feature out there. People entitled to their opinions but thorough and educated ones should matter (much) more.
This is Avigail Idan. She’s a three year old American citizen. She is being held hostage for 45 days and counting by the same monsters who murdered her parents in front of her.
@rooteruditorum@antoniogm Israeli government is defining the policy, IDF only implements it. You have a clear & single address. Israel signed multiple piece treaties in the past incl. with Jordan and Egypt. If you have time on your hands you can learn how these have been implemented and enforced.