←
Palatial352 posts
Palatial avatar

Palatial

@PalatialSim

Physically accurate assets and scenes for robot simulation training at scale

◷ Joined April 2022▧ 102 Photos and videos
Palatial avatar
Palatial@PalatialSim·Nov 10

Dexterity is the last frontier for robotics. Egocentric, dex-umi, and tele-op have so far failed to deliver the quality and scope of data required to solve manipulation. Simulations will. For the past few months, we’ve been working with @origami_robo on the first Dexterity Factory, a real2sim2real pipeline that we believe will solve dexterous manipulation. Objects go in, dexterous policies come out. By combining Palatial’s new real2sim capture system and Mad Max agent with Origami’s S1 Skill-capture gloves and robot hand, we’ve been able to close the sim2real gap on hundreds of complex tasks. We want to contribute this research to the robotics community. Next week, we’ll open source the first 100 hand object interaction dataset, and we’re just getting started.

2:14
Palatial avatar
Palatial@PalatialSim·Oct 22

Nearly 1,000 Palatial generated assets are now part of SimReady Central, NVIDIA's new open library of SimReady OpenUSD assets on Hugging Face. We're excited to contribute to this open ecosystem alongside @standardbots, @LGUS, @FRANKAROBOTICS, @Universal_Robot, and other industry partners, making it easier for developers to build environments for robot training and evaluation. Explore SimReady Central: huggingface.co/spaces/nvidia… Read about our assets: palatial.cloud/updates/open-…

0:58
Palatial avatar
Palatial@PalatialSim·Sep 23

So how does Mad Max compare against Astra? We put them head to head with identical prompts and inputs. Mad Max came out ahead in the following areas: -Geometric accuracy (proportions, scale) -Subtle geometry details (normal maps) -Cleaner Topology -More detailed PBR materials -Fully functional mechanical joints Want to compare yourself? Mad Max is free to try for the next week at dashboard.palatial.cloud (4/4)

0:35
Palatial avatar
Palatial@PalatialSim·Sep 23

Bring your reference materials and context, Mad Max reasons about the task, does research and synthesizes messy input data into faithful recreations of real world objects across use cases and industries. Each asset is validated against its target simulator, ensuring stability, working joints and photoreal materials. (3/4)

0:20
Palatial avatar
Palatial@PalatialSim·Sep 23

Simulation has an N+1 problem. In order to create accurate simulations, there are hundreds of engineering problems to solve and solving one doesn't always lead to solutions for the millions of other scenarios out there. This engineering gap is the primary reason sim is not ubiquitous in robotics today. By combining frontier models like GPT-6 Astra and our own simulation-in-the-loop verification engine, we’ve created general solutions to many long-standing issues in asset generation, including open ended articulation, using an approach that scales with frontier progress. (2/4)

Palatial avatar
Palatial@PalatialSim·Sep 23

Introducing Palatial Mad Max, a state-of-the-art parametric generation agent that unlocks scalable Real2Sim for robotics and physical AI. It's a generalized agent that can create extremely detailed SimReady assets with precise articulated parts and grounded physics properties designed for dexterous manipulation tasks. Try it now at dashboard.palatial.cloud (1/4)

1:19
Palatial avatar
Palatial@PalatialSim·Sep 22

In less than 24 hours, we're unlocking a new tier of capability for robotics simulations. Tune in tomorrow 10am PST

0:29
Palatial avatar
Palatial@PalatialSim·Sep 21

Team has been cooking up a huge update for the past couple of months. SimReady assets are about to change forever. Tune in Wednesday.

0:18
Palatial avatar
Palatial@PalatialSim·Sep 16

Check out some of the cool things folks have already built with the MCP!

My latest kitchen helper has six legs and wings 🪰 Inspired by Google’s fruit-fly brain work, I built a miniature kitchen with @PalatialSim's new MCP in less than an hour, then came roughly 18 hours of training, tuning and testing in MuJoCo to get the fly interacting with it.

1:47
Palatial avatar
Palatial@PalatialSim·Sep 16

Most of the teams we work with live in two places: the simulator and the terminal. This week, we're shipping Palatial into both. Download the @nvidia Isaac Sim plugin to generate new or import existing sim-ready assets directly in the stage. Or, use the MCP and ask your coding agent to call Palatial to create complete simulation packages for Isaac Sim, MuJoCo, or Newton. try it out here: palatial.cloud/isaac-sim-plu…

0:26
Palatial avatar
Palatial@PalatialSim·Sep 2

Our talk at @siggraph 2026: Building SimReady 3D Worlds for Physical AI, is now live. We cover what sim-ready actually means, common pitfalls, and the key challenges we solved to automate asset generation at the scale needed for robotics development. Check it out if you missed it and thanks to the @nvidia team for inviting us!

18:42
Palatial avatar
Palatial@PalatialSim·Sep 1

At @siggraph 2026 we announced the world’s first SimReady asset agent and today we’re making it publicly available in open beta. You can now generate and iterate on assets through natural language. Just upload your reference data, tell us your use case and the agent will figure out the best settings. Try it out now @ dashboard.palatial.cloud This kicks off shiptember. We've got a full month of releases lined up!

0:40
RuiningLi avatar
⟳ Palatial retweeted
RuiningLi@RayLi234·Jun 15

🚀 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: instruct-particulate.github.… 🤗 Demo: huggingface.co/spaces/rayli/…

0:54