Simulation assets for
humanoid robots
Humanoid robots interact with entire environments. Rigyd generates physics-accurate furniture, tools, appliances, and everyday objects for whole-body manipulation and navigation training.
The problem
Why existing workflows fall short.
Humanoids interact with everything
Unlike single-arm robots, humanoids need to navigate, grasp, push, pull, and manipulate objects across entire rooms. Every object needs physics.
Home environments are complex
Training a humanoid for home tasks requires thousands of diverse objects — furniture, kitchen items, tools, electronics — all with accurate physical properties.
Whole-body contact is unforgiving
Humanoid locomotion and manipulation involve full-body contacts. Inaccurate object physics leads to unrealistic fall recovery and interaction behaviors.
How Rigyd helps
AI-native infrastructure that automates the hard parts.
Full environment object sets
Convert entire 3D catalogs of household and industrial objects. Every item gets mass, friction, restitution, and collision geometry for realistic interaction.
Furniture-scale collision meshes
From chairs to refrigerators, Rigyd generates collision geometry that captures functional surfaces — seat shapes, handle positions, shelf clearances.
Semantic labels included
Each asset gets semantic labels and material identification, enabling perception-in-the-loop training and scene understanding tasks.
better real-world transfer with accurate synthetic training
saved per 1,000-object environment build
SimReady assets exist today — humanoids need 10x that
Build environments your humanoid can learn from
Upload household 3D models and get physics-accurate training assets in minutes.
Starts at $29/month. 30 credits included.
Frequently asked questions
Can Rigyd generate assets for whole-body interaction training?
Yes. Rigyd converts entire environment catalogs — furniture, appliances, tools — with mass, collision geometry, and semantic labels suitable for whole-body humanoid manipulation and locomotion tasks.
Does Rigyd support home-scale environment simulation?
Yes. Teams have used Rigyd to convert 1,000+ household and industrial 3D catalogs into SimReady assets. Enterprise API makes ongoing environment expansion practical as new object classes are added.
Are semantic labels included for perception training?
Yes. Every asset ships with semantic labels and material identification so humanoid policies can combine perception and manipulation in the loop — without a separate annotation pass.
Do Rigyd assets work with NVIDIA Isaac GR00T and humanoid foundation models?
Yes. GR00T training pipelines consume USDPhysics-compliant assets via Isaac Sim and Isaac Lab — exactly Rigyd's default output. The 40% real-world performance lift reported in the GR00T N1 benchmark (NVIDIA, 2025) depends on physics-accurate training data, which is the core problem Rigyd solves. Diverse, physics-calibrated objects are what foundation models need to generalize.
Can Rigyd generate full home or kitchen environments, not just individual objects?
Yes. Upload a 3D scan or BIM model of a room and Rigyd processes every distinct object inside it — furniture, appliances, fixtures, decor — into individual SimReady assets with shared scene transforms. The output is a USD scene referencing per-object physics-enabled prims, ready to drop into Isaac Sim for whole-body manipulation training in realistic environments. The per-object structure matters: humanoid policies need to manipulate individual objects (open a drawer, lift a chair, pour from a kettle), not interact with a single room-scale mesh. Rigyd's scene decomposition preserves that interactability automatically, which is otherwise the slowest part of preparing a home-scale simulation environment by hand.
Related reading
In-depth guides on robotics simulation, OpenUSD, and SimReady assets.
21 MAY 2026
Synthetic data generation for computer vision in robotics
Real-world labeled robotics data is expensive and slow to capture. Synthetic data is fast and unlimited — but only useful if the underlying simulation has correct physics, semantic labels, and domain randomization. Here's the complete pipeline.
18 MAY 2026
Digital twin creation pipeline for manufacturing
A factory digital twin needs every object to behave physically, not just render. This is the end-to-end pipeline: CAD intake, BIM merge, physics layer, semantic labeling, simulation runtime — at the asset volumes (10K+ unique SKUs) real factories actually contain.
13 MAY 2026
How to set up mass, friction, and joint properties for robot training
The three pillars of robot physics setup — mass, friction, joints — determine whether your trained policy transfers to real hardware. Here's the calibration target for each, the schemas, and the common mistakes that quietly break training.
Explore more
SimReady for Isaac Sim
OpenUSD assets validated for NVIDIA Isaac Sim
Models for MuJoCo
Physics-accurate assets for robot learning
Assets for Gazebo
Simulation-ready models for ROS 2
Warehouse Simulation
50,000+ SKUs need physics properties
Humanoid Robots
Whole-body interaction training assets
Sim-to-Real Transfer
Close the gap with accurate physics data