A child consumes more data in 1 month than any LLM has ever seen. Embodied agents learn by doing, but the data that teaches them is tactile, sensorial and causal.
Such data does not exist.
To make physical AGI possible, we need to generate this new data at an industrial scale.
Enter Palatial: automated infrastructure that converts raw data into sensory rich playgrounds for robots to learn in.
Today, we’re unveiling Palatial PhysReady, the first automated sim asset generator (try it ⬇️) [1/5]
🚀 Introducing Instruct-Particulate, our new model for inferring articulated structures from static 3D meshes, with significantly improved generalization to novel object categories and support for kinematic prompting.
To achieve this, we scaled our training data 40× and redesigned the model to follow kinematic prompts.
The result: diverse, realistic, simulator-compatible articulated 3D assets can now be generated directly from real-world images!
🔗 Project page: https://t.co/i3GRXPisTo
🤗 Demo: https://t.co/8sGZ8rWh71
@hlynakim Really appreciate that! Getting proper physics and articulation right for manipulation tasks has been such a challenge in sim �� excited to see what you build with it 🤖
The only way to scale robotics is to scale simulation. And sim is bottlenecked by high quality assets.
Today we're releasing Palatial V1.0, the first automated pipeline to generate Sim-Ready assets at scale.
Agent-guided and multimodal, it supports soft-body, articulation, and is the first and only automated asset generator for Nvidia's Newton simulator. @NVIDIARobotics
Start generating: https://t.co/m8iBTXUvpZ
@elodiebear_m We export to multiple formats including OBJ and FBX which give you full geometry access! Our assets are built with proper mesh topology, not just visual shells. You'll get the actual geometry data you need for CAD workflows, not simulation shadows 🛠️
Agent-powered simulation generation.
Soft bodies, powered in Newton (soon integrated into Isaac sim).
Articulated open ended objects, with an interface to improve the outputs.
We are expanding to more input modalities next - think chat for physical AI.
Join us if you want to solve some of the hardest problems in simulations!
We're on a mission to simulate every class of object robotic manipulation can touch, and make it accessible in any simulator.
Over the following weeks we’ll be posting demos, examples, and tutorials. Stay tuned.
https://t.co/u9VhBS2nKB
We're also releasing a Palatial integration layer for Newton: solver and asset params wrapped into clean API schemas for rigid body, cloth, and cable types. Drop Palatial assets straight into a Newton simulation with minimal setup.
https://t.co/BVHPfO91WW
Over the last couple of months we've generated over 10k assets across homes, warehouses, factories, and construction sites. Our community has been building wild things with it:
https://t.co/12zqTAHflp
https://t.co/849wCT3610
https://t.co/mNvXqt5Sgq
Our first Newton release includes support for soft body cables and clothing, as well as rigid and articulated objects.
Generate once, use across NVIDIA Isaac Sim, MuJoCo, and Newton, all Nvidia-approved and validated against the SimReady Foundation standard. @NVIDIARobotics
The pipeline is guided by the Palatial agent: it reads your context, configures the right settings automatically. We accept multimodal inputs: text, images, CAD, or datasheets, just upload and let it do the rest.
Scalable sim data is a foundational bottleneck for robotics. and the building blocks are physically accurate 3D assets.
Today we're releasing the Palatial Library (https://t.co/EHHgKHw7cp), the fastest-growing collection of sim-ready assets for robotics, powered by our automated generation pipeline.
Every 60 seconds, a new Palatial asset is created, 24/7. Our goal is 1 asset/second, and 1 million assets generated this year.