Video wird geladen...

Video konnte nicht geladen werden

Zur Startseite

Introducing ✨RigidFormer: Learning Rigid Dynamics with Transformers - our attempt to scale learning-based physical dynamics with Transformers. RigidFormer learns rigid dynamics with Transformers. It is a mesh-free, object-centric Transformer for multi-object rigid-body contact dynamics from point clouds. Learning physics with purely neural simulators, without relying on traditional physics engines,...

frankzydou's profile picture

Zhiyang (Frank) Dou

2,128 subscribers

573,369 Aufrufe • vor 4 Monaten •via X (Twitter)

BildungWissenschaft & Technologie
LIVE

Anya Rossi• Live Now

Private livecam show

0 Kommentare

Keine Kommentare verfügbar

Kommentare vom Original-Post werden hier angezeigt

Ähnliche Videos

Nvidia announces GAvatar: Animatable 3D Gaussian Avatars with Implicit Mesh Learning paper page: Gaussian splatting has emerged as a powerful 3D representation that harnesses the advantages of both explicit (mesh) and implicit (NeRF) 3D representations. In this paper, we seek to leverage Gaussian splatting to generate realistic animatable avatars from textual descriptions, addressing the limitations (e.g., flexibility and efficiency) imposed by mesh or NeRF-based representations. However, a naive application of Gaussian splatting cannot generate high-quality animatable avatars and suffers from learning instability; it also cannot capture fine avatar geometries and often leads to degenerate body parts. To tackle these problems, we first propose a primitive-based 3D Gaussian representation where Gaussians are defined inside pose-driven primitives to facilitate animation. Second, to stabilize and amortize the learning of millions of Gaussians, we propose to use neural implicit fields to predict the Gaussian attributes (e.g., colors). Finally, to capture fine avatar geometries and extract detailed meshes, we propose a novel SDF-based implicit mesh learning approach for 3D Gaussians that regularizes the underlying geometries and extracts highly detailed textured meshes. Our proposed method, GAvatar, enables the large-scale generation of diverse animatable avatars using only text prompts. GAvatar significantly surpasses existing methods in terms of both appearance and geometry quality, and achieves extremely fast rendering (100 fps) at 1K resolution.
5:36

Nvidia announces GAvatar: Animatable 3D Gaussian Avatars with Implicit Mesh Learning paper page: Gaussian splatting has emerged as a powerful 3D representation that harnesses the advantages of both explicit (mesh) and implicit (NeRF) 3D representations. In this paper, we seek to leverage Gaussian splatting to generate realistic animatable avatars from textual descriptions, addressing the limitations (e.g., flexibility and efficiency) imposed by mesh or NeRF-based representations. However, a naive application of Gaussian splatting cannot generate high-quality animatable avatars and suffers from learning instability; it also cannot capture fine avatar geometries and often leads to degenerate body parts. To tackle these problems, we first propose a primitive-based 3D Gaussian representation where Gaussians are defined inside pose-driven primitives to facilitate animation. Second, to stabilize and amortize the learning of millions of Gaussians, we propose to use neural implicit fields to predict the Gaussian attributes (e.g., colors). Finally, to capture fine avatar geometries and extract detailed meshes, we propose a novel SDF-based implicit mesh learning approach for 3D Gaussians that regularizes the underlying geometries and extracts highly detailed textured meshes. Our proposed method, GAvatar, enables the large-scale generation of diverse animatable avatars using only text prompts. GAvatar significantly surpasses existing methods in terms of both appearance and geometry quality, and achieves extremely fast rendering (100 fps) at 1K resolution.

AK

141,058 Aufrufe • vor 2 Jahren

In my past research experience, finding or developing an appropriate simulation environment, dataset, and benchmark has always been a challenge. Missing features, limited support, or unexpected bugs often occupied my days and nights. Moreover, current simulation platforms are relatively fragmented—making it challenging to replicate the success of the RT-X dataset in unifying community efforts. Introducing RoboVerse, we provide a unified platform, dataset, and benchmark for scalable and generalizable robot learning. We hope to build a shared foundation to combine the community efforts. RoboVerse includes: MetaSim: We carefully designed a configuration system and a universal interface to align current robotic simulators. With MetaSim, you can use any simulator with the same code—bringing together the community’s diverse efforts under one framework! RoboVerse Dataset and Benchmark: We unify popular simulation environments and benchmarks into a single cohesive system and introduce the RoboVerse dataset—a large-scale, high-quality synthetic dataset. Additionally, we propose a standardized benchmark across both imitation learning and reinforcement learning. A cool feature enabled by our unified framework: Hybrid Simulation! You can now integrate physics engines and renderers from different simulators—e.g., using MuJoCo precise physics with Isaac photorealistic rendering. This not only elevates simulation fidelity but also significantly enhances real-world transfer performance across complex robotic applications. Hopefully, our team’s efforts could serve the robotic community to thrive vibrantly in the years to come. RoboVerse is open-sourced🥳!!! Project Page: Documentation: Github Repo: Paper:
2:21

In my past research experience, finding or developing an appropriate simulation environment, dataset, and benchmark has always been a challenge. Missing features, limited support, or unexpected bugs often occupied my days and nights. Moreover, current simulation platforms are relatively fragmented—making it challenging to replicate the success of the RT-X dataset in unifying community efforts. Introducing RoboVerse, we provide a unified platform, dataset, and benchmark for scalable and generalizable robot learning. We hope to build a shared foundation to combine the community efforts. RoboVerse includes: MetaSim: We carefully designed a configuration system and a universal interface to align current robotic simulators. With MetaSim, you can use any simulator with the same code—bringing together the community’s diverse efforts under one framework! RoboVerse Dataset and Benchmark: We unify popular simulation environments and benchmarks into a single cohesive system and introduce the RoboVerse dataset—a large-scale, high-quality synthetic dataset. Additionally, we propose a standardized benchmark across both imitation learning and reinforcement learning. A cool feature enabled by our unified framework: Hybrid Simulation! You can now integrate physics engines and renderers from different simulators—e.g., using MuJoCo precise physics with Isaac photorealistic rendering. This not only elevates simulation fidelity but also significantly enhances real-world transfer performance across complex robotic applications. Hopefully, our team’s efforts could serve the robotic community to thrive vibrantly in the years to come. RoboVerse is open-sourced🥳!!! Project Page: Documentation: Github Repo: Paper:

Haoran Geng

84,318 Aufrufe • vor 1 Jahr

We are back again :) After three weeks of quiet building. Introducing Genesis World 1.0, our latest simulation platform, the second release in our full-stack suite. Open-sourced. Robotics is still bottlenecked by the 1× speed of the physical world. Every model, checkpoint, and data recipe eventually needs to be tested on physical hardware, slowly, expensively, and with limited coverage. One hour in reality can become 100 days in simulation. That is how robotics model iteration moves from a wall-clock bottleneck to a compute problem. To make this work, simulation has to be both fast and trustworthy. Over the past year, we rebuilt the entire stack: a GPU-accelerated cross-platform compiler, penetration-free multi-physics contact solvers, unified rigid and deformable physics, and a photo-realistic renderer purpose-built for physical AI applications. We built Nyx, a high-performance path-traced rendering engine for robotics application. Genesis World 1.0 achieves near realtime performance with our latest development for penetration-free IPC solver, supporting various types of deformables beyond rigid bodies. It supports contact-rich, dexterous manipulation simulation across different embodiments: unitree, sharpa, wuji, genesis hand and various types of grippers. Under the hood is Quadrants, our effort in pushing forward cross-platform GPU-accelerated computation. Quadrants started as a fork of Taichi, and we rebuilt most of the critical parts for optimizing simulation workloads, giving 10x faster launch time and up to 4.6x runtime performance compared to the initial Genesis release. Together, they bring us to an unprecedentedly low sim-to-real gap, enabling zero-shot real-to-sim model evaluation and much faster iteration of GENE. All available today. Genesis World 1.0: Quadrants: Nyx:
2:51

We are back again :) After three weeks of quiet building. Introducing Genesis World 1.0, our latest simulation platform, the second release in our full-stack suite. Open-sourced. Robotics is still bottlenecked by the 1× speed of the physical world. Every model, checkpoint, and data recipe eventually needs to be tested on physical hardware, slowly, expensively, and with limited coverage. One hour in reality can become 100 days in simulation. That is how robotics model iteration moves from a wall-clock bottleneck to a compute problem. To make this work, simulation has to be both fast and trustworthy. Over the past year, we rebuilt the entire stack: a GPU-accelerated cross-platform compiler, penetration-free multi-physics contact solvers, unified rigid and deformable physics, and a photo-realistic renderer purpose-built for physical AI applications. We built Nyx, a high-performance path-traced rendering engine for robotics application. Genesis World 1.0 achieves near realtime performance with our latest development for penetration-free IPC solver, supporting various types of deformables beyond rigid bodies. It supports contact-rich, dexterous manipulation simulation across different embodiments: unitree, sharpa, wuji, genesis hand and various types of grippers. Under the hood is Quadrants, our effort in pushing forward cross-platform GPU-accelerated computation. Quadrants started as a fork of Taichi, and we rebuilt most of the critical parts for optimizing simulation workloads, giving 10x faster launch time and up to 4.6x runtime performance compared to the initial Genesis release. Together, they bring us to an unprecedentedly low sim-to-real gap, enabling zero-shot real-to-sim model evaluation and much faster iteration of GENE. All available today. Genesis World 1.0: Quadrants: Nyx:

Genesis AI

317,293 Aufrufe • vor 3 Monaten

Special thanks to Google DeepMind for inviting me to try out Genie 3. I'm excited to share my thoughts on this early research prototype and also some of my live recordings below: I spent the whole day playing with the system and when it works, it is truly mind blowing🤯. It is the first neural game engine / world model I have tried that generalizes so well and has long term world consistency. Here’s a couple of examples from my live recording and some thoughts on what it means for the future of gaming, robotics, digital experiences and ASI. Where it shines: - Truly general-purpose and quick startup time. Works exceptionally well for gaming environments but also generalizes to other industrial and real-world scenarios. - It learns physics. Although there are systematic failures even for rigid body physics, it was clear to me that it can learn game engine and non-rigid physics without an underlying engine (and in limit learn from game engines via training data). - It works exceptionally well for stylized environments with characters walking around. This will have implications for concept artists, level designers and game devs. - It is way more fun than video models, indicating that there are high retention consumer experiences waiting to be built with this in the future - Photorealistic walk throughs and drone shots work exceptionally well - Global illumination and lighting works surprisingly well - Visual memory is quite powerful and the same objects approximately remain coherent under occlusion and longer time horizons Open Problems: - Physics is still hard and there are obvious failure cases when I tried the classical intuitive physics experiments from psychology (tower of blocks). - Social and multi-agent interactions are tricky to handle. 1vs1 combat games do not work - Long instruction following and simple combinatorial game logic fails (e.g. collect some points / keys etc, go to the door, unlock and so on) - Action space is limited - It is far from being a real game engines and has a long way to go but this is a clear glimpse into the future. The Future: - It is impressive enough for me to have strong conviction that this is going to disrupt the gaming industry. It is super early days and there are a lot of failures but the writing is on the wall. Lots of challenging scientific, engineering and scaling problems to be solved but it is going to happen in the next 5 years. - This is the final piece before we get full AGI and now I think we are well on our way to truly solve it once something like this is scaled up. In many ways it is more ASI than AGI but this is a matter of definitions. The fidelity and generalizability will reach human-level and quickly surpass humans - People are going to combine this with 3D AI and LLMs to build AAA games.
0:57

Special thanks to Google DeepMind for inviting me to try out Genie 3. I'm excited to share my thoughts on this early research prototype and also some of my live recordings below: I spent the whole day playing with the system and when it works, it is truly mind blowing🤯. It is the first neural game engine / world model I have tried that generalizes so well and has long term world consistency. Here’s a couple of examples from my live recording and some thoughts on what it means for the future of gaming, robotics, digital experiences and ASI. Where it shines: - Truly general-purpose and quick startup time. Works exceptionally well for gaming environments but also generalizes to other industrial and real-world scenarios. - It learns physics. Although there are systematic failures even for rigid body physics, it was clear to me that it can learn game engine and non-rigid physics without an underlying engine (and in limit learn from game engines via training data). - It works exceptionally well for stylized environments with characters walking around. This will have implications for concept artists, level designers and game devs. - It is way more fun than video models, indicating that there are high retention consumer experiences waiting to be built with this in the future - Photorealistic walk throughs and drone shots work exceptionally well - Global illumination and lighting works surprisingly well - Visual memory is quite powerful and the same objects approximately remain coherent under occlusion and longer time horizons Open Problems: - Physics is still hard and there are obvious failure cases when I tried the classical intuitive physics experiments from psychology (tower of blocks). - Social and multi-agent interactions are tricky to handle. 1vs1 combat games do not work - Long instruction following and simple combinatorial game logic fails (e.g. collect some points / keys etc, go to the door, unlock and so on) - Action space is limited - It is far from being a real game engines and has a long way to go but this is a clear glimpse into the future. The Future: - It is impressive enough for me to have strong conviction that this is going to disrupt the gaming industry. It is super early days and there are a lot of failures but the writing is on the wall. Lots of challenging scientific, engineering and scaling problems to be solved but it is going to happen in the next 5 years. - This is the final piece before we get full AGI and now I think we are well on our way to truly solve it once something like this is scaled up. In many ways it is more ASI than AGI but this is a matter of definitions. The fidelity and generalizability will reach human-level and quickly surpass humans - People are going to combine this with 3D AI and LLMs to build AAA games.

Tejas Kulkarni

88,083 Aufrufe • vor 1 Jahr

This is probably the most entertaining way to understand one of AI’s hardest AI debates. Transformer vs Post-Transformer, argued by leading researchers, inside a real physical boxing ring. Both technically deep and genuinely entertaining. I was glued for the entire 1 hour 20 minutes. So many super cool points to learn. 🥊 Transformers - Transformers still own the present because they work at scale. They are simple, trainable, hardware-friendly, and already power the strongest AI systems we use today. - The Transformer is basically a memory machine. It stores information as keys and values, then uses attention to pull back the most useful parts when answering. - The real Transformer advantage is not just “attention.” The bigger advantage is that it fits modern hardware extremely well, so it can process huge batches of tokens fast. - Scaling is still the brutal rule. If you give Transformers more compute, more data, and more parameters, they usually keep getting better. Any Post-Transformer architecture has to scale just as well, or better. - It is not enough to look clever on small tests, because the real question is whether it improves faster than Transformers when scaled up. - A replacement cannot be slightly better. Because the whole AI stack is already built around Transformers, the next architecture may need to be around 10x better to force everyone to switch. - Transformers are powerful, but they may be brute force. A human does not need to read the entire internet many times to become smart, but current LLMs need enormous data and compute. 🥊 Post-Transformer - Post-Transformer people are not saying Transformers are bad. They are saying Transformers may be the best current tool, not the final form of machine intelligence. - The biggest Post-Transformer target is native reasoning and continual learning. Today’s LLM reasoning often feels like text-based step-by-step work added on top, instead of thinking happening naturally inside the model. - Latent reasoning is one possible next step. That means the model reasons inside its own hidden internal space, instead of writing every thought out as words. - Continual learning is still a major weakness. Humans keep learning from experience, but most Transformer-based models are trained, frozen, and then only adapt inside the prompt. - Long context is not the same as real memory. A model can read a huge prompt, but that is different from building a life history, learning from mistakes, and updating beliefs over time. - The future may be hybrid, not a clean replacement. Transformers may stay as 1 building block while newer systems add better memory, better reasoning, and better learning loops. - The most interesting possibility is that Transformers may help discover their own successor. AI agents are already getting better at research and coding, so the next architecture may come from AI-assisted architecture search. ------- - Benchmarks are a problem. Many public benchmarks are easy to game, so they may show leaderboard strength without proving deeper intelligence. - Perplexity is still probably a great metric to evaluate frontier models,, because it tests prediction quality. --- Overall, Transformers continue to dominate, but the frontier is clearly widening. Pathway’s BDH (Dragon Hatchling — brain-inspired reasoning architecture), Sakana AI’s CTMs (Continuous Thought Machines — models that think over time), and Liquid AI’s LFMs (Liquid Foundation Models — efficient multimodal foundation models) - all of these show how the frontier is expanding. --- From “Pathway (pathway[.]com)” Youtube channel (link in comment) Zuzanna Stamirowska
2:35

This is probably the most entertaining way to understand one of AI’s hardest AI debates. Transformer vs Post-Transformer, argued by leading researchers, inside a real physical boxing ring. Both technically deep and genuinely entertaining. I was glued for the entire 1 hour 20 minutes. So many super cool points to learn. 🥊 Transformers - Transformers still own the present because they work at scale. They are simple, trainable, hardware-friendly, and already power the strongest AI systems we use today. - The Transformer is basically a memory machine. It stores information as keys and values, then uses attention to pull back the most useful parts when answering. - The real Transformer advantage is not just “attention.” The bigger advantage is that it fits modern hardware extremely well, so it can process huge batches of tokens fast. - Scaling is still the brutal rule. If you give Transformers more compute, more data, and more parameters, they usually keep getting better. Any Post-Transformer architecture has to scale just as well, or better. - It is not enough to look clever on small tests, because the real question is whether it improves faster than Transformers when scaled up. - A replacement cannot be slightly better. Because the whole AI stack is already built around Transformers, the next architecture may need to be around 10x better to force everyone to switch. - Transformers are powerful, but they may be brute force. A human does not need to read the entire internet many times to become smart, but current LLMs need enormous data and compute. 🥊 Post-Transformer - Post-Transformer people are not saying Transformers are bad. They are saying Transformers may be the best current tool, not the final form of machine intelligence. - The biggest Post-Transformer target is native reasoning and continual learning. Today’s LLM reasoning often feels like text-based step-by-step work added on top, instead of thinking happening naturally inside the model. - Latent reasoning is one possible next step. That means the model reasons inside its own hidden internal space, instead of writing every thought out as words. - Continual learning is still a major weakness. Humans keep learning from experience, but most Transformer-based models are trained, frozen, and then only adapt inside the prompt. - Long context is not the same as real memory. A model can read a huge prompt, but that is different from building a life history, learning from mistakes, and updating beliefs over time. - The future may be hybrid, not a clean replacement. Transformers may stay as 1 building block while newer systems add better memory, better reasoning, and better learning loops. - The most interesting possibility is that Transformers may help discover their own successor. AI agents are already getting better at research and coding, so the next architecture may come from AI-assisted architecture search. ------- - Benchmarks are a problem. Many public benchmarks are easy to game, so they may show leaderboard strength without proving deeper intelligence. - Perplexity is still probably a great metric to evaluate frontier models,, because it tests prediction quality. --- Overall, Transformers continue to dominate, but the frontier is clearly widening. Pathway’s BDH (Dragon Hatchling — brain-inspired reasoning architecture), Sakana AI’s CTMs (Continuous Thought Machines — models that think over time), and Liquid AI’s LFMs (Liquid Foundation Models — efficient multimodal foundation models) - all of these show how the frontier is expanding. --- From “Pathway (pathway[.]com)” Youtube channel (link in comment) Zuzanna Stamirowska

Rohan Paul

89,110 Aufrufe • vor 3 Monaten

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.
0:40

New Paper: Continuous Thought Machines 🧠 Neurons in brains use timing and synchronization in the way that they compute, but this is largely ignored in modern neural nets. We believe neural timing is key for the flexibility and adaptability of biological intelligence. We propose a new neural architecture, “Continuous Thought Machines” (CTMs), which is built from the ground up to use neural dynamics as a core representation for intelligence. By using neural dynamics as a first-class representational citizen, CTMs naturally perform adaptive computation. Many emergent, interesting behaviors arise as a result: CTMs solve mazes by observing a raw maze image and producing step-by-step instructions directly from its neural dynamics. When tasked with image recognition, the CTM naturally takes multiple steps to examine different parts of the image before making its decision. This step-by-step approach not only makes its behavior more interpretable but also improves accuracy: the longer it “thinks,” the more accurate its answers become. We also found that this allows the CTM to decide to spend less time thinking on simpler images, thus saving energy. When identifying a gorilla, for example, the CTM’s attention moves from eyes to nose to mouth in a pattern remarkably similar to human visual attention. I think this work underscores an important, yet often lost, synergy between neuroscience and AI. While modern AI is ostensibly brain-inspired, the two fields often operate in surprising isolation. By starting with such inspiration and iteratively following the emergent, interesting behaviors, we developed a model with unexpected capabilities, such as its surprisingly strong calibration in classification tasks, a feature that was not explicitly designed for. When we initially asked, “why do this research?”, we hoped the journey of the CTM would provide compelling answers. By embracing light biological inspiration and pursuing the novel behaviors observed, we have arrived at a model with emergent capabilities that exceeded our initial designs. We are committed to continuing this exploration, borrowing further concepts to discover what new and exciting behaviors will emerge, pushing the boundaries of what AI can achieve.

hardmaru

257,577 Aufrufe • vor 1 Jahr

Vibe Coding 3D Garment Software with ThreeJS : A Small Step For Me So, after modeling the human i did what any reasonable vibe coder would do, i asked codex how to get clothes for my models After it was done running subliminal ad campaigns for Marvelous Designer and CLO 3D, i asked it to explain their architecture to me and adapt it to my threejs app. Guess what it did? You damn right, it built the most basic shit interpretation you can think of. And this is the average interaction the Anti-AI coders have until they conclude that AI is slop and/or it can only work if you micro manage it on every line of code. Well, eons of humanities knowledge are now packaged in tiny silicon and transferred across the globe in realtime, available on tap. So anyways i just iterated quite a lot over it, told it repeatedly why it was bad (the initial one used rapier physics and a naive cloth simulation) We found out together that: 1. A ground truth document model is needed 2. The visual mesh in 3D should be triangulated from the 2D shape 3. The physical object is running independently through different solvers: - A fast proxy which is generated by reading all the bones in runtime and just inflating these areas with spheres and capsules - A medium quality proxy which resamples the human model and creates a lower-poly mesh for simulations - Full mesh simulation (can't run it, every simulation tick takes about 5 minutes on my machine) It ended the session by telling me that this is still crap because it runs everything on CPU (thanks, not that i care, but i guess we'll be fixing that?) Oh yea also built a 2D canvas editor with boolean operations so i can build cool stuff like ponchos. It also allows me to mark stitches between two objects, which is how the shirt in the video pulls towards the other half. The garment's properties and materials are not yet exposed, yes i know it looks very stiff like a poncho made from a persian rug, we're working on it, okay? So, yea, tbh this is another endless rabbit hole, let's go i guess
0:27

Vibe Coding 3D Garment Software with ThreeJS : A Small Step For Me So, after modeling the human i did what any reasonable vibe coder would do, i asked codex how to get clothes for my models After it was done running subliminal ad campaigns for Marvelous Designer and CLO 3D, i asked it to explain their architecture to me and adapt it to my threejs app. Guess what it did? You damn right, it built the most basic shit interpretation you can think of. And this is the average interaction the Anti-AI coders have until they conclude that AI is slop and/or it can only work if you micro manage it on every line of code. Well, eons of humanities knowledge are now packaged in tiny silicon and transferred across the globe in realtime, available on tap. So anyways i just iterated quite a lot over it, told it repeatedly why it was bad (the initial one used rapier physics and a naive cloth simulation) We found out together that: 1. A ground truth document model is needed 2. The visual mesh in 3D should be triangulated from the 2D shape 3. The physical object is running independently through different solvers: - A fast proxy which is generated by reading all the bones in runtime and just inflating these areas with spheres and capsules - A medium quality proxy which resamples the human model and creates a lower-poly mesh for simulations - Full mesh simulation (can't run it, every simulation tick takes about 5 minutes on my machine) It ended the session by telling me that this is still crap because it runs everything on CPU (thanks, not that i care, but i guess we'll be fixing that?) Oh yea also built a 2D canvas editor with boolean operations so i can build cool stuff like ponchos. It also allows me to mark stitches between two objects, which is how the shirt in the video pulls towards the other half. The garment's properties and materials are not yet exposed, yes i know it looks very stiff like a poncho made from a persian rug, we're working on it, okay? So, yea, tbh this is another endless rabbit hole, let's go i guess

robot

38,935 Aufrufe • vor 3 Monaten

Physics-based Motion Retargeting from Sparse Inputs paper page: Avatars are important to create interactive and immersive experiences in virtual worlds. One challenge in animating these characters to mimic a user's motion is that commercial AR/VR products consist only of a headset and controllers, providing very limited sensor data of the user's pose. Another challenge is that an avatar might have a different skeleton structure than a human and the mapping between them is unclear. In this work we address both of these challenges. We introduce a method to retarget motions in real-time from sparse human sensor data to characters of various morphologies. Our method uses reinforcement learning to train a policy to control characters in a physics simulator. We only require human motion capture data for training, without relying on artist-generated animations for each avatar. This allows us to use large motion capture datasets to train general policies that can track unseen users from real and sparse data in real-time. We demonstrate the feasibility of our approach on three characters with different skeleton structure: a dinosaur, a mouse-like creature and a human. We show that the avatar poses often match the user surprisingly well, despite having no sensor information of the lower body available. We discuss and ablate the important components in our framework, specifically the kinematic retargeting step, the imitation, contact and action reward as well as our asymmetric actor-critic observations. We further explore the robustness of our method in a variety of settings including unbalancing, dancing and sports motions.
7:12

Physics-based Motion Retargeting from Sparse Inputs paper page: Avatars are important to create interactive and immersive experiences in virtual worlds. One challenge in animating these characters to mimic a user's motion is that commercial AR/VR products consist only of a headset and controllers, providing very limited sensor data of the user's pose. Another challenge is that an avatar might have a different skeleton structure than a human and the mapping between them is unclear. In this work we address both of these challenges. We introduce a method to retarget motions in real-time from sparse human sensor data to characters of various morphologies. Our method uses reinforcement learning to train a policy to control characters in a physics simulator. We only require human motion capture data for training, without relying on artist-generated animations for each avatar. This allows us to use large motion capture datasets to train general policies that can track unseen users from real and sparse data in real-time. We demonstrate the feasibility of our approach on three characters with different skeleton structure: a dinosaur, a mouse-like creature and a human. We show that the avatar poses often match the user surprisingly well, despite having no sensor information of the lower body available. We discuss and ablate the important components in our framework, specifically the kinematic retargeting step, the imitation, contact and action reward as well as our asymmetric actor-critic observations. We further explore the robustness of our method in a variety of settings including unbalancing, dancing and sports motions.

AK

106,527 Aufrufe • vor 3 Jahren

I’m thrilled to announce that we just released GraspGen, a multi-year project we have been cooking at NVIDIA Robotics 🚀 GraspGen: A Diffusion-Based Framework for 6-DOF Grasping Grasping is a foundational challenge in robotics 🤖 — whether for industrial picking or general-purpose humanoids. VLA + real data collection is all the rage now but is expensive and scales poorly for this task. For every new gripper and/or scene, you’ll have to recollect the dataset in this paradigm for the best perf. 💡Key Idea: Since grasping is such a well-defined task in simulation - why can’t we just scale synthetic data generation and train a generative model for grasping? By embracing modularity and standardized grasp formats, we can make this a turnkey technology that works zero-shot for multiple settings. GraspGen is a modular framework for diffusion-based 6-DOF grasp generation that scales across embodiment types, observability conditions, clutter, task complexity. Key Features: ✅ Multi-embodiment support: suction, parallel-jaw, and multi-fingered grippers ✅ Generalization to partial + complete 3D point clouds ✅ Generalization to single-objects + cluttered scenes ✅ Modular design uses other robotics modules and foundation models (SAM2, cuRobo, FoundationStereo, FoundationPose). This allows GraspGen to focus on only one thing - grasp generation ✅ Training recipe: grasp discriminator is trained with On-Generator data from the diffusion model - so that it learns to correct the mistakes (if any) of the diffusion generator ✅ Real-time performance (~20 Hz) before any GPU acceleration; low memory footprint 📊 Results: • SOTA on the FetchBench [Han et al. CoRL 2024] benchmark • Zero-shot sim-to-real transfer on unknown objects and cluttered scenes • Dataset of 53M simulated grasps across 8K objects from Objaverse 📄 arXiv: 🌐 Website: 💻 Code: A huge thank you to everyone involved in this journey — excited to see what the community builds on top of it! Joint work with Clemens Eppner , Balakumar Sundaralingam , Yu-Wei, Jun Yamada Wentao Yuan and other collaborators #robotics #diffusionmodels #physicalAI #simtoreal
0:39

I’m thrilled to announce that we just released GraspGen, a multi-year project we have been cooking at NVIDIA Robotics 🚀 GraspGen: A Diffusion-Based Framework for 6-DOF Grasping Grasping is a foundational challenge in robotics 🤖 — whether for industrial picking or general-purpose humanoids. VLA + real data collection is all the rage now but is expensive and scales poorly for this task. For every new gripper and/or scene, you’ll have to recollect the dataset in this paradigm for the best perf. 💡Key Idea: Since grasping is such a well-defined task in simulation - why can’t we just scale synthetic data generation and train a generative model for grasping? By embracing modularity and standardized grasp formats, we can make this a turnkey technology that works zero-shot for multiple settings. GraspGen is a modular framework for diffusion-based 6-DOF grasp generation that scales across embodiment types, observability conditions, clutter, task complexity. Key Features: ✅ Multi-embodiment support: suction, parallel-jaw, and multi-fingered grippers ✅ Generalization to partial + complete 3D point clouds ✅ Generalization to single-objects + cluttered scenes ✅ Modular design uses other robotics modules and foundation models (SAM2, cuRobo, FoundationStereo, FoundationPose). This allows GraspGen to focus on only one thing - grasp generation ✅ Training recipe: grasp discriminator is trained with On-Generator data from the diffusion model - so that it learns to correct the mistakes (if any) of the diffusion generator ✅ Real-time performance (~20 Hz) before any GPU acceleration; low memory footprint 📊 Results: • SOTA on the FetchBench [Han et al. CoRL 2024] benchmark • Zero-shot sim-to-real transfer on unknown objects and cluttered scenes • Dataset of 53M simulated grasps across 8K objects from Objaverse 📄 arXiv: 🌐 Website: 💻 Code: A huge thank you to everyone involved in this journey — excited to see what the community builds on top of it! Joint work with Clemens Eppner , Balakumar Sundaralingam , Yu-Wei, Jun Yamada Wentao Yuan and other collaborators #robotics #diffusionmodels #physicalAI #simtoreal

Adithya Murali

24,296 Aufrufe • vor 1 Jahr

Secure attachment during infancy lays the foundation for our lifelong mental and physical health. As I wind down this week’s series on infant/toddler attachment, I wanted to conclude with a few final words about why it matters so much. Obviously, strong relational connections between parents and children are beneficial in and of themselves. But there’s more to it than that. Attachment grounds and shapes our long-term mental health as well as our ability to enter into positive relationships with others. You see, children who are securely attached tend to be just that: Secure. This allows them to navigate the world independently, with confidence in themselves and their abilities. It almost seems contradictory that independence would spring from dependence. But it most surely does. The more love and care children receive during their earliest months and years and the more attentively and predictably their needs are met by treasured loved ones, the more they experience the world as safe place. And this is a gift to both the brain and the body. When the human body is not flooded daily with stress hormones and when anxiety occurs - not as a constant state - but more occasionally in response to life’s challenges, not only are we better prepared to deal with it - but our bodies don’t bear the long-term burden of a constant stress response. This is a fascinating area of early development that we are still only beginning to fully understand, but suffice it to say that positive early experiences - both in the home and in conjunction with high quality early education and care programs- are increasingly linked to lifelong health. This brain-body connection has huge implications not only for individual children, but for our public policy. A tweet is hardly the place to unpack this conversation in full, but know that my new book, The Daycare Myth (which can be ordered from the link in my bio) explores in detail the disconnect between what we know about the science of early development and how it is enacted (or, more frequently, not enacted) in our public policy. The costs - both developmentally and economically - are astronomical. And not only to individuals, but to taxpayers - who would be far better served by making small investments in getting things right in the first place than paying for the results of our inaction across the lifespan. It all begins with attachment. Thanks for following along this week! ——— This lovely video - which, to be clear, is NOT an illustration of insecure attachment - was shared to IG by olliejamesfaulds.
0:20

Secure attachment during infancy lays the foundation for our lifelong mental and physical health. As I wind down this week’s series on infant/toddler attachment, I wanted to conclude with a few final words about why it matters so much. Obviously, strong relational connections between parents and children are beneficial in and of themselves. But there’s more to it than that. Attachment grounds and shapes our long-term mental health as well as our ability to enter into positive relationships with others. You see, children who are securely attached tend to be just that: Secure. This allows them to navigate the world independently, with confidence in themselves and their abilities. It almost seems contradictory that independence would spring from dependence. But it most surely does. The more love and care children receive during their earliest months and years and the more attentively and predictably their needs are met by treasured loved ones, the more they experience the world as safe place. And this is a gift to both the brain and the body. When the human body is not flooded daily with stress hormones and when anxiety occurs - not as a constant state - but more occasionally in response to life’s challenges, not only are we better prepared to deal with it - but our bodies don’t bear the long-term burden of a constant stress response. This is a fascinating area of early development that we are still only beginning to fully understand, but suffice it to say that positive early experiences - both in the home and in conjunction with high quality early education and care programs- are increasingly linked to lifelong health. This brain-body connection has huge implications not only for individual children, but for our public policy. A tweet is hardly the place to unpack this conversation in full, but know that my new book, The Daycare Myth (which can be ordered from the link in my bio) explores in detail the disconnect between what we know about the science of early development and how it is enacted (or, more frequently, not enacted) in our public policy. The costs - both developmentally and economically - are astronomical. And not only to individuals, but to taxpayers - who would be far better served by making small investments in getting things right in the first place than paying for the results of our inaction across the lifespan. It all begins with attachment. Thanks for following along this week! ——— This lovely video - which, to be clear, is NOT an illustration of insecure attachment - was shared to IG by olliejamesfaulds.

Dan Wuori

101,524 Aufrufe • vor 2 Jahren

How does an embryo reliably "compute" its form - "cell by cell" - using only local interactions and mechanics, yet produce a precise global body plan? I’m excited to share our Nature Methods paper "MultiCell: geometric learning in multicellular development", presenting #AIxBiology research led by Haiqian Yang and the result of a great collaboration with Ming Guo, George Roy, Tomer Stern, Anh Nguyen and Dapeng Bi. A long-standing challenge in developmental biology is to predict how thousands of cells collectively self-organize as tissues fold, divide, and rearrange. In MultiCell, we represent a developing embryo as a dual graph that unifies two complementary views of tissue mechanics with single-cell resolution: cells as moving points (granular) and cells as a connected foam (junction network). This lets the model learn dynamics from both geometry and cell–cell connectivity. On whole-embryo 4D light-sheet movies of Drosophila gastrulation (~5,000 cells), our model predicts key cell behaviors and the timing of events, including junction loss, rearrangements, and divisions with high accuracy, at single-cell resolution. Beyond prediction, the same representation supports robust time alignment across embryos and offers interpretable activation maps that highlight the morphogenetic "drivers" of development. The broader goal is a foundation for cell-by-cell forecasting in more complex tissues, and eventually for detecting subtle dynamical signatures of disease. Kudos to the team for this inspiring collaboration with brilliant researchers to push the boundary of AI for biology! Citation: Yang, H., Roy, G., Nguyen, A.Q., Buehler, M.J., et al. MultiCell: geometric learning in multicellular development. Nature Methods (2025), DOI: 10.1038/s41592-025-02983-x Code/data links are in the manuscript.
0:46

How does an embryo reliably "compute" its form - "cell by cell" - using only local interactions and mechanics, yet produce a precise global body plan? I’m excited to share our Nature Methods paper "MultiCell: geometric learning in multicellular development", presenting #AIxBiology research led by Haiqian Yang and the result of a great collaboration with Ming Guo, George Roy, Tomer Stern, Anh Nguyen and Dapeng Bi. A long-standing challenge in developmental biology is to predict how thousands of cells collectively self-organize as tissues fold, divide, and rearrange. In MultiCell, we represent a developing embryo as a dual graph that unifies two complementary views of tissue mechanics with single-cell resolution: cells as moving points (granular) and cells as a connected foam (junction network). This lets the model learn dynamics from both geometry and cell–cell connectivity. On whole-embryo 4D light-sheet movies of Drosophila gastrulation (~5,000 cells), our model predicts key cell behaviors and the timing of events, including junction loss, rearrangements, and divisions with high accuracy, at single-cell resolution. Beyond prediction, the same representation supports robust time alignment across embryos and offers interpretable activation maps that highlight the morphogenetic "drivers" of development. The broader goal is a foundation for cell-by-cell forecasting in more complex tissues, and eventually for detecting subtle dynamical signatures of disease. Kudos to the team for this inspiring collaboration with brilliant researchers to push the boundary of AI for biology! Citation: Yang, H., Roy, G., Nguyen, A.Q., Buehler, M.J., et al. MultiCell: geometric learning in multicellular development. Nature Methods (2025), DOI: 10.1038/s41592-025-02983-x Code/data links are in the manuscript.

Markus J. Buehler

388,303 Aufrufe • vor 8 Monaten

Can GPT-4 teach a robot hand to do pen spinning tricks better than you do? I'm excited to announce Eureka, an open-ended agent that designs reward functions for robot dexterity at super-human level. It’s like Voyager in the space of a physics simulator API! Eureka bridges the gap between high-level reasoning (coding) and low-level motor control. It is a “hybrid-gradient architecture”: a black box, inference-only LLM instructs a white box, learnable neural network. The outer loop runs GPT-4 to refine the reward function (gradient-free), while the inner loop runs reinforcement learning to train a robot controller (gradient-based). We are able to scale up Eureka thanks to IsaacGym, a GPU-accelerated physics simulator that speeds up reality by 1000x. On a benchmark suite of 29 tasks across 10 robots, Eureka rewards outperform expert human-written ones on 83% of the tasks by 52% improvement margin on average. We are surprised that Eureka is able to learn pen spinning tricks, which are very difficult even for CGI artists to animate frame by frame! Eureka also enables a new form of in-context RLHF, which is able to incorporate a human operator’s feedback in natural language to steer and align the reward functions. It can serve as a powerful co-pilot for robot engineers to design sophisticated motor behaviors. As usual, we open-source everything! Welcome you all to check out our video gallery and try the codebase today: Paper: Code: Deep dive with me: 🧵
0:24

Can GPT-4 teach a robot hand to do pen spinning tricks better than you do? I'm excited to announce Eureka, an open-ended agent that designs reward functions for robot dexterity at super-human level. It’s like Voyager in the space of a physics simulator API! Eureka bridges the gap between high-level reasoning (coding) and low-level motor control. It is a “hybrid-gradient architecture”: a black box, inference-only LLM instructs a white box, learnable neural network. The outer loop runs GPT-4 to refine the reward function (gradient-free), while the inner loop runs reinforcement learning to train a robot controller (gradient-based). We are able to scale up Eureka thanks to IsaacGym, a GPU-accelerated physics simulator that speeds up reality by 1000x. On a benchmark suite of 29 tasks across 10 robots, Eureka rewards outperform expert human-written ones on 83% of the tasks by 52% improvement margin on average. We are surprised that Eureka is able to learn pen spinning tricks, which are very difficult even for CGI artists to animate frame by frame! Eureka also enables a new form of in-context RLHF, which is able to incorporate a human operator’s feedback in natural language to steer and align the reward functions. It can serve as a powerful co-pilot for robot engineers to design sophisticated motor behaviors. As usual, we open-source everything! Welcome you all to check out our video gallery and try the codebase today: Paper: Code: Deep dive with me: 🧵

Jim Fan

2,674,581 Aufrufe • vor 2 Jahren

A transformer can learn not just the outcomes of dynamics, but the operator that executes the rules. To show this we trained a transformer on roughly 0.04% of a discrete rule space - 100 of 262,144 possible rules - and it learned to apply unseen rules from the same rule class. The model does not simply memorize specific rules. It learns the operator that maps a supplied rule plus an initial state, including unseen rules from this class, to the correct next state. This is relevant because it is a shift from “neural networks approximate dynamics” to “neural networks can learn to execute symbolic programs within a defined rule class”. The rule itself is supplied at inference time, as data, and the network has internalized how rules act, not which rules to apply. On previously unseen rules, the model achieves 98.5% perfect one-step forecasts and reconstructs governing rules with up to 96% functional accuracy. Two results make this hold up under scrutiny. First, inductive bias decay. As we scaled training rule diversity, the correlation between functional inference accuracy and distance-from-nearest-training-rule collapsed to R² = 0.00. At the largest tested training-rule diversity, the model’s performance on a new rule shows no measurable dependence on how similar that rule is to anything it was trained on. The bias toward training data (the thing we worry most about in compositional generalization claims) is something we can measure decaying, and we find that at scale it is gone. Second, an identifiability theory. We derive a closed-form expression for the number of rules consistent with a single observation. This reframes the inverse problem: failure to recover ground truth is not necessarily a model defect, but can be correct behavior when the data underdetermine the rule. The model is sampling the equivalence class; and identifiability is governed by coverage, not capacity. The methodological move underneath both results is amortization. Classical work on rule inference (e.g. the Santa Fe EVCA program, evolutionary search over CA rule space) was per-instance: search the rule space for each new system. We replace that with a single forward pass of a transformer trained across many instantiations of the rule class. That is what makes symbolic rule inference scalable as a research direction rather than a curiosity. We show that this works in a tightly constrained domain: binary, deterministic, local cellular automata on small grids. The locality-break experiment shows the model fails sharply when target systems violate its structural priors (which is itself a useful diagnostic, but it bounds the operator class). We don't yet know how this scales to multistate, higher-dimensional, or stochastic CA, or whether it transfers cleanly to non-CA systems whose coarse-grained dynamics admit local surrogates. The identifiability framework - what can be inferred from observation, given a hypothesis class - should transfer wherever finite local rules meet sparse data. The amortization argument transfers wherever per-instance symbolic search has been the bottleneck. Those are the pieces I expect to outlive the cellular automata setting. Led by Jaime Berkovich with Noah David, at LAMM@MIT. Out now in Advanced Science Advanced Portfolio (link to paper & code below).
0:49

A transformer can learn not just the outcomes of dynamics, but the operator that executes the rules. To show this we trained a transformer on roughly 0.04% of a discrete rule space - 100 of 262,144 possible rules - and it learned to apply unseen rules from the same rule class. The model does not simply memorize specific rules. It learns the operator that maps a supplied rule plus an initial state, including unseen rules from this class, to the correct next state. This is relevant because it is a shift from “neural networks approximate dynamics” to “neural networks can learn to execute symbolic programs within a defined rule class”. The rule itself is supplied at inference time, as data, and the network has internalized how rules act, not which rules to apply. On previously unseen rules, the model achieves 98.5% perfect one-step forecasts and reconstructs governing rules with up to 96% functional accuracy. Two results make this hold up under scrutiny. First, inductive bias decay. As we scaled training rule diversity, the correlation between functional inference accuracy and distance-from-nearest-training-rule collapsed to R² = 0.00. At the largest tested training-rule diversity, the model’s performance on a new rule shows no measurable dependence on how similar that rule is to anything it was trained on. The bias toward training data (the thing we worry most about in compositional generalization claims) is something we can measure decaying, and we find that at scale it is gone. Second, an identifiability theory. We derive a closed-form expression for the number of rules consistent with a single observation. This reframes the inverse problem: failure to recover ground truth is not necessarily a model defect, but can be correct behavior when the data underdetermine the rule. The model is sampling the equivalence class; and identifiability is governed by coverage, not capacity. The methodological move underneath both results is amortization. Classical work on rule inference (e.g. the Santa Fe EVCA program, evolutionary search over CA rule space) was per-instance: search the rule space for each new system. We replace that with a single forward pass of a transformer trained across many instantiations of the rule class. That is what makes symbolic rule inference scalable as a research direction rather than a curiosity. We show that this works in a tightly constrained domain: binary, deterministic, local cellular automata on small grids. The locality-break experiment shows the model fails sharply when target systems violate its structural priors (which is itself a useful diagnostic, but it bounds the operator class). We don't yet know how this scales to multistate, higher-dimensional, or stochastic CA, or whether it transfers cleanly to non-CA systems whose coarse-grained dynamics admit local surrogates. The identifiability framework - what can be inferred from observation, given a hypothesis class - should transfer wherever finite local rules meet sparse data. The amortization argument transfers wherever per-instance symbolic search has been the bottleneck. Those are the pieces I expect to outlive the cellular automata setting. Led by Jaime Berkovich with Noah David, at LAMM@MIT. Out now in Advanced Science Advanced Portfolio (link to paper & code below).

Markus J. Buehler

39,019 Aufrufe • vor 4 Monaten

Yann LeCun (Yann LeCun ) beautifully explains how the architecture and principles used to train LLMs can not be extended to teach AI the real-world intelligence. In 1 line: LLMs excel where intelligence equals sequence prediction over symbols. Real-world intelligence requires learned world models, abstraction, causality, and action planning under uncertainty, which current next-token training does not provide. He says current LLMs learn by predicting the next token. That objective works very well when the task itself can be reduced to manipulating discrete symbols and sequences. Math, physics problem solving on paper, and coding fit this pattern because success largely comes from searching and composing the right sequences of symbols, equations, or program tokens. With enough data and scale, these models get very good at that kind of structured sequence prediction. Real-world intelligence is different. The physical world is continuous, noisy, uncertain, and high dimensional. To act in it, a system needs internal models that capture objects, dynamics, causality, constraints from the body, and the outcomes of actions over time. Humans and animals build abstract representations from rich sensory streams, then make predictions in that abstract space, not at the raw pixel level. That is why a child can learn intuitive physics, plan multi-step actions, and adapt quickly in new situations with little data. His claim about saturation follows from this gap. Scaling token prediction keeps improving symbol manipulation tasks like math and code, but it hits limits on embodied reasoning and common sense because text alone does not provide the right learning signals for world models. Predicting the next word cannot efficiently teach contact forces, affordances, occlusion, friction, or how actions change the state of the environment. For that, he argues we need architectures that learn abstractions from sensory data and predict futures in abstract latent spaces, then use those predictions to plan actions toward goals with built-in guardrails. --- From 'Pioneer Works' YT Channel (link in comment)
1:13

Yann LeCun (Yann LeCun ) beautifully explains how the architecture and principles used to train LLMs can not be extended to teach AI the real-world intelligence. In 1 line: LLMs excel where intelligence equals sequence prediction over symbols. Real-world intelligence requires learned world models, abstraction, causality, and action planning under uncertainty, which current next-token training does not provide. He says current LLMs learn by predicting the next token. That objective works very well when the task itself can be reduced to manipulating discrete symbols and sequences. Math, physics problem solving on paper, and coding fit this pattern because success largely comes from searching and composing the right sequences of symbols, equations, or program tokens. With enough data and scale, these models get very good at that kind of structured sequence prediction. Real-world intelligence is different. The physical world is continuous, noisy, uncertain, and high dimensional. To act in it, a system needs internal models that capture objects, dynamics, causality, constraints from the body, and the outcomes of actions over time. Humans and animals build abstract representations from rich sensory streams, then make predictions in that abstract space, not at the raw pixel level. That is why a child can learn intuitive physics, plan multi-step actions, and adapt quickly in new situations with little data. His claim about saturation follows from this gap. Scaling token prediction keeps improving symbol manipulation tasks like math and code, but it hits limits on embodied reasoning and common sense because text alone does not provide the right learning signals for world models. Predicting the next word cannot efficiently teach contact forces, affordances, occlusion, friction, or how actions change the state of the environment. For that, he argues we need architectures that learn abstractions from sensory data and predict futures in abstract latent spaces, then use those predictions to plan actions toward goals with built-in guardrails. --- From 'Pioneer Works' YT Channel (link in comment)

Rohan Paul

104,460 Aufrufe • vor 8 Monaten

Here is the Geometry Nodes Weighted Normals with Laplacian Blur on a full character (a vroid). It easily improves the shading even on game topology with almost no setup. I built this as part of my quest to improve real time toon shading. 3D anime models are popular, but use of dynamic light is rare even among high quality vtuber models. This is for several reasons, but a big one is simply that it takes a lot of Custom Normals work to make 3D cel shading not look like a jagged mess (other pieces of the puzzle are issues like deformations, multiple lights, etc). And fixing Normals is tedious, especially on existing game topology. I have focused on proxy meshes for priority areas like character faces, but they aren't an efficient solution for the whole body + outfit. I wanted something I could just throw on any model and make it at least not a jagged mess anymore even if it wasn't perfect. As you can see from this clip, this does that very well! And vertex groups can be used to control the style of the effect and power. It still can't smooth beyond what the topology density can support, but the topo itself is no longer a problem (for higher res, could be run on a subdivided version of the model and then baked to a Normal Map.) The only changes I made to this model were adding a weld modifier to merge split edges during interpolation, and a vertex group to select the skirt. I have not yet added full handling and logic for detecting edges with big angles like the skirt, or for handling boundaries like on the hair, so both those areas can get better too. You can also see that while it successfully smooths out the Face, it isn't really stylistically correct there. That is still best done with a proxy mesh to define a new shape. This is part of the tools I am working on for Fondant. We are putting together a Blender Addon to release this + a proxy mesh tool for the face, and are working on resolving other problems in-engine to fully bring dynamic light to real time 3D toon shading. Give us a follow, and send them a DM if you are interested in testing these tools as they develop!
2:17

Here is the Geometry Nodes Weighted Normals with Laplacian Blur on a full character (a vroid). It easily improves the shading even on game topology with almost no setup. I built this as part of my quest to improve real time toon shading. 3D anime models are popular, but use of dynamic light is rare even among high quality vtuber models. This is for several reasons, but a big one is simply that it takes a lot of Custom Normals work to make 3D cel shading not look like a jagged mess (other pieces of the puzzle are issues like deformations, multiple lights, etc). And fixing Normals is tedious, especially on existing game topology. I have focused on proxy meshes for priority areas like character faces, but they aren't an efficient solution for the whole body + outfit. I wanted something I could just throw on any model and make it at least not a jagged mess anymore even if it wasn't perfect. As you can see from this clip, this does that very well! And vertex groups can be used to control the style of the effect and power. It still can't smooth beyond what the topology density can support, but the topo itself is no longer a problem (for higher res, could be run on a subdivided version of the model and then baked to a Normal Map.) The only changes I made to this model were adding a weld modifier to merge split edges during interpolation, and a vertex group to select the skirt. I have not yet added full handling and logic for detecting edges with big angles like the skirt, or for handling boundaries like on the hair, so both those areas can get better too. You can also see that while it successfully smooths out the Face, it isn't really stylistically correct there. That is still best done with a proxy mesh to define a new shape. This is part of the tools I am working on for Fondant. We are putting together a Blender Addon to release this + a proxy mesh tool for the face, and are working on resolving other problems in-engine to fully bring dynamic light to real time 3D toon shading. Give us a follow, and send them a DM if you are interested in testing these tools as they develop!

aVersionOfReality

14,904 Aufrufe • vor 1 Jahr

Exciting updates on Project GR00T! We discover a systematic way to scale up robot data, tackling the most painful pain point in robotics. The idea is simple: human collects demonstration on a real robot, and we multiply that data 1000x or more in simulation. Let’s break it down: 1. We use Apple Vision Pro (yes!!) to give the human operator first person control of the humanoid. Vision Pro parses human hand pose and retargets the motion to the robot hand, all in real time. From the human’s point of view, they are immersed in another body like the Avatar. Teleoperation is slow and time-consuming, but we can afford to collect a small amount of data. 2. We use RoboCasa, a generative simulation framework, to multiply the demonstration data by varying the visual appearance and layout of the environment. In Jensen’s keynote video below, the humanoid is now placing the cup in hundreds of kitchens with a huge diversity of textures, furniture, and object placement. We only have 1 physical kitchen at the GEAR Lab in NVIDIA HQ, but we can conjure up infinite ones in simulation. 3. Finally, we apply MimicGen, a technique to multiply the above data even more by varying the *motion* of the robot. MimicGen generates vast number of new action trajectories based on the original human data, and filters out failed ones (e.g. those that drop the cup) to form a much larger dataset. To sum up, given 1 human trajectory with Vision Pro -> RoboCasa produces N (varying visuals) -> MimicGen further augments to NxM (varying motions). This is the way to trade compute for expensive human data by GPU-accelerated simulation. A while ago, I mentioned that teleoperation is fundamentally not scalable, because we are always limited by 24 hrs/robot/day in the world of atoms. Our new GR00T synthetic data pipeline breaks this barrier in the world of bits. Scaling has been so much fun for LLMs, and it's finally our turn to have fun in robotics! We are building tools to enable everyone in the ecosystem to scale up with us. Links in thread:
1:21

Exciting updates on Project GR00T! We discover a systematic way to scale up robot data, tackling the most painful pain point in robotics. The idea is simple: human collects demonstration on a real robot, and we multiply that data 1000x or more in simulation. Let’s break it down: 1. We use Apple Vision Pro (yes!!) to give the human operator first person control of the humanoid. Vision Pro parses human hand pose and retargets the motion to the robot hand, all in real time. From the human’s point of view, they are immersed in another body like the Avatar. Teleoperation is slow and time-consuming, but we can afford to collect a small amount of data. 2. We use RoboCasa, a generative simulation framework, to multiply the demonstration data by varying the visual appearance and layout of the environment. In Jensen’s keynote video below, the humanoid is now placing the cup in hundreds of kitchens with a huge diversity of textures, furniture, and object placement. We only have 1 physical kitchen at the GEAR Lab in NVIDIA HQ, but we can conjure up infinite ones in simulation. 3. Finally, we apply MimicGen, a technique to multiply the above data even more by varying the *motion* of the robot. MimicGen generates vast number of new action trajectories based on the original human data, and filters out failed ones (e.g. those that drop the cup) to form a much larger dataset. To sum up, given 1 human trajectory with Vision Pro -> RoboCasa produces N (varying visuals) -> MimicGen further augments to NxM (varying motions). This is the way to trade compute for expensive human data by GPU-accelerated simulation. A while ago, I mentioned that teleoperation is fundamentally not scalable, because we are always limited by 24 hrs/robot/day in the world of atoms. Our new GR00T synthetic data pipeline breaks this barrier in the world of bits. Scaling has been so much fun for LLMs, and it's finally our turn to have fun in robotics! We are building tools to enable everyone in the ecosystem to scale up with us. Links in thread:

Jim Fan

364,565 Aufrufe • vor 2 Jahren

Can we compile matter - for instance, a pine cone - and derive new active materials, end-to-end from observation to manufacturing? If physical systems can be formalized as composable mathematics, we can point AI that has been shown to resolve long-open mathematical problems at matter itself. Our new work turns bioinspired engineering from analogy into formal compilation: biology and mechanics become explicit, checkable, and executable, so AI reasoning can produce physical designs. This is the first end-to-end demonstration in which a formally compositional multiscale model is carried from a biological hierarchy, through engineered design and fabrication specification, to executable manufacturing code - and then to a physically tested artifact. Background: Humans have long been inspired by biology to advance technology, but this has usually been an ad hoc process rather than a mathematically rigorous one. Natural materials such as pinecones achieve adaptive behavior through mechanisms organized across many scales. Engineering typically translates those mechanisms by analogy: identify a biological principle, build something inspired by it, and validate each new design as a separate case. This can produce remarkable results, but the knowledge does not readily compound. Instead, we represent each scale as a dynamical module with explicit states, stimuli, governing laws, and interfaces. Every scale-to-scale map must preserve the stimulus - response dynamics: evolve the fine-scale system and then map upward, or map upward first and then evolve. The two paths must agree. Because this condition is preserved under composition, locally valid interfaces remain consistent when assembled into the full hierarchy. We then carry that structure into an engineered system, translate the target behavior into a verified fabrication specification, and compile it into G-code: the toolpaths, deposition sequence, temperatures, speeds, and other commands executed by a 3D printer. The intermediate translations are explicit, checkable, and executable rather than completed through an ad hoc handoff. The formal guarantee is that given valid local models and interfaces, their composition remains valid. Whether those models and manufacturing assumptions accurately capture physical reality remains an empirical question. That is why we fabricated and tested the results. We generated four actuator classes by crossing two stimuli - humidity and heat - with two responses: bending and twisting. The fourth, thermal twisting, required no new pipeline and no separate derivation within the framework. It emerged by composing a thermal stimulus module already validated in one case with a twisting module validated in another. The generated G-code produced the intended motion without manual redesign, and all four predictions fell within one experimental standard deviation of the measured response. Why this matters: 1⃣For AI in science, this provides a physics-aware type system against which generative proposals can be checked - and rejected at the interface - before expensive simulation, fabrication, or experiment. It is roughly analogous to proof checking, but for the composition of physical mechanisms. 2⃣For engineering, the accessible design space can scale with a library of validated components rather than with the number of individually derived cases. 3⃣The mathematics, category theory, carries all the way into a physical object on a print bed. This points toward scientific knowledge as executable infrastructure: models that are not only described in papers, but typed, composable, verifiable, and able to compile into experiments. Excellent work led by my student Lee Marom with Skylar Tibbits & Gioele Zardini. Paper published in J. Mech. Phys. Solids along with code, Grasshopper scripts, and manufacturing G-code below.
1:21

Can we compile matter - for instance, a pine cone - and derive new active materials, end-to-end from observation to manufacturing? If physical systems can be formalized as composable mathematics, we can point AI that has been shown to resolve long-open mathematical problems at matter itself. Our new work turns bioinspired engineering from analogy into formal compilation: biology and mechanics become explicit, checkable, and executable, so AI reasoning can produce physical designs. This is the first end-to-end demonstration in which a formally compositional multiscale model is carried from a biological hierarchy, through engineered design and fabrication specification, to executable manufacturing code - and then to a physically tested artifact. Background: Humans have long been inspired by biology to advance technology, but this has usually been an ad hoc process rather than a mathematically rigorous one. Natural materials such as pinecones achieve adaptive behavior through mechanisms organized across many scales. Engineering typically translates those mechanisms by analogy: identify a biological principle, build something inspired by it, and validate each new design as a separate case. This can produce remarkable results, but the knowledge does not readily compound. Instead, we represent each scale as a dynamical module with explicit states, stimuli, governing laws, and interfaces. Every scale-to-scale map must preserve the stimulus - response dynamics: evolve the fine-scale system and then map upward, or map upward first and then evolve. The two paths must agree. Because this condition is preserved under composition, locally valid interfaces remain consistent when assembled into the full hierarchy. We then carry that structure into an engineered system, translate the target behavior into a verified fabrication specification, and compile it into G-code: the toolpaths, deposition sequence, temperatures, speeds, and other commands executed by a 3D printer. The intermediate translations are explicit, checkable, and executable rather than completed through an ad hoc handoff. The formal guarantee is that given valid local models and interfaces, their composition remains valid. Whether those models and manufacturing assumptions accurately capture physical reality remains an empirical question. That is why we fabricated and tested the results. We generated four actuator classes by crossing two stimuli - humidity and heat - with two responses: bending and twisting. The fourth, thermal twisting, required no new pipeline and no separate derivation within the framework. It emerged by composing a thermal stimulus module already validated in one case with a twisting module validated in another. The generated G-code produced the intended motion without manual redesign, and all four predictions fell within one experimental standard deviation of the measured response. Why this matters: 1⃣For AI in science, this provides a physics-aware type system against which generative proposals can be checked - and rejected at the interface - before expensive simulation, fabrication, or experiment. It is roughly analogous to proof checking, but for the composition of physical mechanisms. 2⃣For engineering, the accessible design space can scale with a library of validated components rather than with the number of individually derived cases. 3⃣The mathematics, category theory, carries all the way into a physical object on a print bed. This points toward scientific knowledge as executable infrastructure: models that are not only described in papers, but typed, composable, verifiable, and able to compile into experiments. Excellent work led by my student Lee Marom with Skylar Tibbits & Gioele Zardini. Paper published in J. Mech. Phys. Solids along with code, Grasshopper scripts, and manufacturing G-code below.

Markus J. Buehler

128,390 Aufrufe • vor 1 Monat

Great updates today. It has been sad the level of attacks and attempts to be-little what ProffittHarsad Community and dev Team had built and had to endure in order to contribute a solution and a product for everyone. We do use Jupiter quote APIs and Jupiter on-chain Routers. Right now they are the ONLY Transmission/Engines available on chain for us to build the Experience for our users while we build in the coming (weeks) our Engine and own Transmission. Lamborghini is not bad cause it used other manufacturers parts while learning how to build the best for their needs. Red Bull Is not a CAR manufacturer, so you can still add value on top of Engine/Transmissions. Its called "PURPOSE & GOALS" We love Jupiter - We do integrate a lot of their amazing open-source APIs in order to make our aggregation experience smooth. We do not iFRAME Jupiter Front End or Websites. There is no shame on this and its intended for the construction of new products and solutions. We also love @heliuslabs as they serve our RPC calls and demands. We also love Amazon Web Services for providing us the Cloud service we need. We also love Solana for the amazing blockchain we all can use. We also love Tensor ⚡️ cause they were one of my inspirations on making the choice to give it a try and Lead this attempt. We love Uniswap 1inch PancakeSwap For showing the WORLD we all could swap in freedom and peace. We are currently building our Front End/Backend Experience, using our servers, our RPCs, our own code to make this happen. We charge a Fee as intended in our ethos for the experience and service we provide. We will continue to improve the experience tailored for our Community and users. We are not interested in Feuds or Comparative battles based on ill intentions. The next stage for is integrate our own Router and wrapper, not cause Jupiter has bad code or solutions, just because we would love to be independent and build our own journey. I have been an avid user of Jupiter for almost 3 years, traded over 8 M USD across all my accounts. Nothing but respects for what they've achieved. Spiderswap is here to build an aggregation solution for everyone. So as someone said "Something like Jupiter but more Community owned". We believe there is a place in which Jupiter and ProffittHarsad can coexist and add value to each other. I personally thank many others for behind scenes support, candor and wisdom. Thanks for helping! With love and Amity! - There is no big vs little. Its just one that already has built and another who started 70 days ago. Itsy Bitsy... 🕷️🤍🕸️
0:24

Great updates today. It has been sad the level of attacks and attempts to be-little what ProffittHarsad Community and dev Team had built and had to endure in order to contribute a solution and a product for everyone. We do use Jupiter quote APIs and Jupiter on-chain Routers. Right now they are the ONLY Transmission/Engines available on chain for us to build the Experience for our users while we build in the coming (weeks) our Engine and own Transmission. Lamborghini is not bad cause it used other manufacturers parts while learning how to build the best for their needs. Red Bull Is not a CAR manufacturer, so you can still add value on top of Engine/Transmissions. Its called "PURPOSE & GOALS" We love Jupiter - We do integrate a lot of their amazing open-source APIs in order to make our aggregation experience smooth. We do not iFRAME Jupiter Front End or Websites. There is no shame on this and its intended for the construction of new products and solutions. We also love @heliuslabs as they serve our RPC calls and demands. We also love Amazon Web Services for providing us the Cloud service we need. We also love Solana for the amazing blockchain we all can use. We also love Tensor ⚡️ cause they were one of my inspirations on making the choice to give it a try and Lead this attempt. We love Uniswap 1inch PancakeSwap For showing the WORLD we all could swap in freedom and peace. We are currently building our Front End/Backend Experience, using our servers, our RPCs, our own code to make this happen. We charge a Fee as intended in our ethos for the experience and service we provide. We will continue to improve the experience tailored for our Community and users. We are not interested in Feuds or Comparative battles based on ill intentions. The next stage for is integrate our own Router and wrapper, not cause Jupiter has bad code or solutions, just because we would love to be independent and build our own journey. I have been an avid user of Jupiter for almost 3 years, traded over 8 M USD across all my accounts. Nothing but respects for what they've achieved. Spiderswap is here to build an aggregation solution for everyone. So as someone said "Something like Jupiter but more Community owned". We believe there is a place in which Jupiter and ProffittHarsad can coexist and add value to each other. I personally thank many others for behind scenes support, candor and wisdom. Thanks for helping! With love and Amity! - There is no big vs little. Its just one that already has built and another who started 70 days ago. Itsy Bitsy... 🕷️🤍🕸️

modsiw

25,540 Aufrufe • vor 2 Jahren

If an AI can control 1,000 robots to perform 1 million skills in 1 billion different simulations, then it may "just work" in our real world, which is simply another point in the vast space of possible realities. This is the fundamental principle behind why simulation works so effectively for robotics. Real-world teleoperation data scales linearly with human time (< 24 hrs/robot/day). Sim data scales exponentially with compute. There are 3 big trends for simulators in the near future: 1. Massive parallelization on large clusters. Physics equations are "just" matrix math at their core. I hear GPUs are good at matrix math 🔥. One can run 100K copies of simulation on a single GPU. To put this number in perspective: 1 hour of wallclock compute time gives a robot 10 years (!!) of training experience. That's how Neo was able to learn martial arts in a blink of an eye in the Matrix Dojo. 2. Generative graphics pipeline. Traditionally, simulators require a huge amount of manual effort from artists: 3D assets, textures, scene layouts, etc. But every component in the workflow can be automated: text-to-image, text-to-3D mesh, and LLMs that write Universal Scene Description (USD) files as a coding exercise. RoboCasa is one example of a prior work. 3. End2end neural net that acts as simulator itself. This is still bluesky research and quite far from replacing a graphics pipeline, but we are seeing some exciting signs-of-life based on video gen models: Sora, Veo2, CogVideoX, Hunyuan (text-to-video); and action-driven world models: GameNGen, Oasis, Genie-2, etc. Genesis does great on (1) for certain tasks, shows good promises on (2), and could become a data generation tool for reaching (3). Its sim2real capabilities for locomotion are good, but there's still a long way to go for contact-rich, dexterous manipulation. It shows a bold vision and is on the right path to providing a virtual cradle for embodied AI. It is open-source and puts a streamlined user journey at the front and center. I had the privilege to know Zhou Xian and play a small part in his project since a year ago. Xian has been crunching code non-stop on Genesis with a very small group of core devs. He often replied to my messages at 3 am. Zhenjia Xu from our GEAR team helped with sim2real experiments in his spare time. Genesis is truly a grassroot effort with an intense focus on quality engineering. Nothing gives me more joy than seeing the simulation ecosystem bloom. Robotics should be a moonshot initiative owned by all of humanity. Congratulations.
3:26

If an AI can control 1,000 robots to perform 1 million skills in 1 billion different simulations, then it may "just work" in our real world, which is simply another point in the vast space of possible realities. This is the fundamental principle behind why simulation works so effectively for robotics. Real-world teleoperation data scales linearly with human time (< 24 hrs/robot/day). Sim data scales exponentially with compute. There are 3 big trends for simulators in the near future: 1. Massive parallelization on large clusters. Physics equations are "just" matrix math at their core. I hear GPUs are good at matrix math 🔥. One can run 100K copies of simulation on a single GPU. To put this number in perspective: 1 hour of wallclock compute time gives a robot 10 years (!!) of training experience. That's how Neo was able to learn martial arts in a blink of an eye in the Matrix Dojo. 2. Generative graphics pipeline. Traditionally, simulators require a huge amount of manual effort from artists: 3D assets, textures, scene layouts, etc. But every component in the workflow can be automated: text-to-image, text-to-3D mesh, and LLMs that write Universal Scene Description (USD) files as a coding exercise. RoboCasa is one example of a prior work. 3. End2end neural net that acts as simulator itself. This is still bluesky research and quite far from replacing a graphics pipeline, but we are seeing some exciting signs-of-life based on video gen models: Sora, Veo2, CogVideoX, Hunyuan (text-to-video); and action-driven world models: GameNGen, Oasis, Genie-2, etc. Genesis does great on (1) for certain tasks, shows good promises on (2), and could become a data generation tool for reaching (3). Its sim2real capabilities for locomotion are good, but there's still a long way to go for contact-rich, dexterous manipulation. It shows a bold vision and is on the right path to providing a virtual cradle for embodied AI. It is open-source and puts a streamlined user journey at the front and center. I had the privilege to know Zhou Xian and play a small part in his project since a year ago. Xian has been crunching code non-stop on Genesis with a very small group of core devs. He often replied to my messages at 3 am. Zhenjia Xu from our GEAR team helped with sim2real experiments in his spare time. Genesis is truly a grassroot effort with an intense focus on quality engineering. Nothing gives me more joy than seeing the simulation ecosystem bloom. Robotics should be a moonshot initiative owned by all of humanity. Congratulations.

Jim Fan

157,343 Aufrufe • vor 1 Jahr

I'm observing a mini Moravec's paradox within robotics: gymnastics that are difficult for humans are much easier for robots than "unsexy" tasks like cooking, cleaning, and assembling. It leads to a cognitive dissonance for people outside the field, "so, robots can parkour & breakdance, but why can't they take care of my dog?" Trust me, I got asked by my parents about this more than you think ... The "Robot Moravec's paradox" also creates the illusion that physical AI capabilities are way more advanced than they truly are. I'm not singling out Unitree, as it applies widely to all recent acrobatic demos in the industry. Here's a simple test: if you set up a wall in front of the side-flipping robot, it will slam into it at full force and make a spectacle. Because it's just overfitting that single reference motion, without any awareness of the surroundings. Here's why the paradox exists: it's much easier to train a "blind gymnast" than a robot that sees and manipulates. The former can be solved entirely in simulation and transferred zero-shot to the real world, while the latter demands extremely realistic rendering, contact physics, and messy real-world object dynamics - none of which can be simulated well. Imagine you can train LLMs not from the internet, but from a purely hand-crafted text console game. Roboticists got lucky. We happen to live in a world where accelerated physics engines are so good that we can get away with impressive acrobatics using literally zero real data. But we haven't yet discovered the same cheat code for general dexterity. Till then, we'll still get questioned by our confused parents.
1:15

I'm observing a mini Moravec's paradox within robotics: gymnastics that are difficult for humans are much easier for robots than "unsexy" tasks like cooking, cleaning, and assembling. It leads to a cognitive dissonance for people outside the field, "so, robots can parkour & breakdance, but why can't they take care of my dog?" Trust me, I got asked by my parents about this more than you think ... The "Robot Moravec's paradox" also creates the illusion that physical AI capabilities are way more advanced than they truly are. I'm not singling out Unitree, as it applies widely to all recent acrobatic demos in the industry. Here's a simple test: if you set up a wall in front of the side-flipping robot, it will slam into it at full force and make a spectacle. Because it's just overfitting that single reference motion, without any awareness of the surroundings. Here's why the paradox exists: it's much easier to train a "blind gymnast" than a robot that sees and manipulates. The former can be solved entirely in simulation and transferred zero-shot to the real world, while the latter demands extremely realistic rendering, contact physics, and messy real-world object dynamics - none of which can be simulated well. Imagine you can train LLMs not from the internet, but from a purely hand-crafted text console game. Roboticists got lucky. We happen to live in a world where accelerated physics engines are so good that we can get away with impressive acrobatics using literally zero real data. But we haven't yet discovered the same cheat code for general dexterity. Till then, we'll still get questioned by our confused parents.

Jim Fan

398,361 Aufrufe • vor 1 Jahr