tipatat

1/N Long horizon, complex tasks that truly matter in everyday life are not solved problems by today’s robotics, requiring planning, object detection, object manipulation, and failure recovery. That's why Stanford's BEHAVIOR Challenge is back for year 2! Last year, the winning solution reached only 12.4% full task success. This year, the BEHAVIOR challenge has more tasks, better evaluation, and is easier to use. 🚨 ⏰ Submission deadline: 10/16/2026 📣 Winners announced: 11/04/2026 🏆 Prize pool: $11,000

⚡️ Tencent HY-World 2.1 is HERE! Not a video. A world. 🪐 3 months just after 2.0, we've supercharged everything: ✨ Cleaner geometry ✨ Sharper rendering ✨ Larger explorable range It's a REAL 3D world you can Walk through and Interact with.

The wildest allegations in Apple’s trade secrets lawsuit against OpenAI techcrunch.com/2026/07/13/the…

So... remember when we joked about making a desktop pet? Yeah, we actually did it. Stay: Ember's Desktop Adventures is now live on Steam and we'd love it if you'd check it out and throw us a wishlist. store.steampowered.com/app/4766080/St… #VirtualPet #DesktopPet #DigitalPet

Lenovo Reportedly Shutters US-based XR Business Unit Amid Shift to AI Wearables See more 👉roadtovr.com/lenovo-shutter…

In sports, the decisive shot or move happens in a fraction of a second. Even with modern broadcasting and replay technology, it's not always possible to see exactly what the athlete did, from every angle and at any speed. This is where 4D Gaussian Splatting comes in: footage captured on the field, in the ring, or anywhere else can be turned into a highly accurate reconstruction, letting you experience sports in a new way. In the video: the swings of Kyle Berkshire, Wyndham Clark, Max Homa, and Min Woo Lee, captured at the PGA Championship by @radiantimages , streamed over @TMobileBusiness onsite 5G network, and reconstructed by Gracia.

🚀 LingBot-VA 2.0 is here! After half a year of teamwork, we're thrilled to release LingBot-VA 2.0 — a native video-action foundation model for generalizable robot control. Unlike prior world-action models that retrofit generic video generators for robot control, LingBot-VA 2.0 is natively pretrained from scratch as a video-action foundation model. Three key insights: 🌍 Native Video-Action Pretraining for learning world knowledge that enables strong generalization. 🧩 Semantic Visual-Action Tokenizer for more accurate action prediction with robust prompt following ⚡ Foresight Reasoning enables the robot to think ahead while acting, delivering continuous, responsive control without interrupting execution. It runs in real time on consumer-grade GPUs, supports up to 150 Hz control, and generalizes to unseen tasks. 👇 Demos below

Bring Marble worlds into Unreal Engine🤘🔥 We partnered with the World Labs community to create two tutorials covering the Gaussian splat workflow in Unreal Engine using Volinga + Akiya plugins. They cover import, collision, relighting, and gameplay setup. Check them out ↓

Everybody is asking me how to get #GaussianSplatting in #VR. This is my answer 👇 You might not like it, but the truth is that right now Unity is not ready and it doesn't matter how hard you try as soon as you go above 400K, the framerate on standalone headsets is terrible. You should instead look at the amazing work that teams at @theworldlabs and @playcanvas are doing. They are pushing innovation and are building stunning tech demos where features like the LOD turns into gameplay mechanic. Using your hands as a sort of torch to define the LOD is genius! This is the best moment to experiment and show innovation. For a second, take a step back from your training and simulations, train an AI on how to build on Playcanvas or Spark and break things.

We’re launching the first in a series of Embodied Reasoning offerings: Perceptron Egocentric. It achieves SOTA over the best robotic annotation pipelines built on Gemini 3.5 Flash and Gemini Robotics-ER 1.6. Early access is available to partners.

Seriously impressive engineering. Vertically integrated and designed from the ground up to be safe, precise, quick, and strong. This dedication to solving the hardest problems first and tackling safety through core innovation are key reasons why I'm at 1x.tech

Introducing 📢📢ProxyPose📢📢 ✨Track 6-DoF motion of any query pixel via video-to-video translation 📰ArXiv: arxiv.org/abs/2607.06555 💻Code: github.com/ruihangzhang97… 🌐Webpage: ruihangzhang97.github.io/proxypose

I have been trying to find something meaningful to say about the Id Software layoffs. My “Microsoft will probably be a good steward of the brand” statement isn’t aging well, and this is certainly going to dampen the mood of the founder reunion at QuakeCon next month. I’m saddened, but I can’t muster anger or outrage over it. I don’t have access to the books, but I suspect that Id Software was a marginal business from Microsoft’s perspective. I believe the reports that Minecraft revenues have been carrying several other studios. To continue being produced long term, games need to succeed, not just be beloved. Games are competing with every other option for spending your leisure time and money, and the competition is brutal. You can’t rule out the possibility that executives are idiots, but that shouldn’t be your default belief. I don’t think there is any obvious path that would have doubled the revenue from Id games. Could they have gotten more with a different pricing strategy? Could they have created more things for fans to buy? Could they have cost effectively marketed in a way that reached more players that would have loved and bought the games? Could they have changed the game designs and broadened the appeal to more players without alienating existing ones? Could they have produced the games at a lower cost, faster or cheaper? I really don’t know. The game isn’t over yet, and I hope the studio rallies through.

4DGS with 75 8K RED Cameras. Done by Prism AI, @Aina, Deckard Render, and FutureDeluxe. Deep dive with the full creative team coming next week.

The future of robotics is something you have to experience hands-on. 🤲 Join us at #SIGGRAPH2026 for our physical AI labs covering NVIDIA Cosmos, robot learning with Isaac Lab and Newton, and more. Check the comments for the full lineup. ⬇️

Today we launched Server Authority as an option for creators. Server Authority helps solve the complex synchronization and physics challenges required to power competitive, hyper-performant multiplayer experiences at massive scale.

Meet VLA-JEPA, the first world model policy in LeRobot 🌍 Most VLAs just learn observation → action. VLA-JEPA also learns action-relevant dynamics: during training, a JEPA world model has to predict upcoming frames in latent space from the model's own actions. Then, at inference, the world model disappears entirely, leaving a standard, fast VLA: Qwen backbone, action head, nothing extra. In our demo it was fine-tuned on just 13 examples and ran in real time on an NVIDIA DGX Spark. On a consumer RTX 3080, it hits 10Hz using under 6GB of VRAM. World-model supervision during training, zero extra cost at inference. That's the whole trick.

Introducing NEO’s 25 Degrees of Freedom, tendon-driven hands — nearing or surpassing human-level dexterity, strength, speed, and reliability. For seventy years, robotics worked around the hand problem. The humanoid bet is the reverse: it lives or dies at the fingertips.

We evaluated 30+ frontier embodied AI models. The result is clear: current generalist robot policies are still far from robust real-world manipulation. This is why we built RoboDojo.