SimReady assets for
Genesis simulator
Genesis is the open-source GPU-native robotics simulator. Drop in 3D models, images, or text descriptions and get MJCF (Genesis-native) or OpenUSD outputs with calibrated mass, friction, restitution, and collision geometry.
The problem
Why existing workflows fall short.
Asset prep is the wall on training throughput
Genesis offers GPU-parallel simulation orders of magnitude faster than CPU simulators. But asset prep, hand-authoring physics, collision meshes, and inertia, still takes ~4 engineer-hours per asset, capping the asset diversity your training can actually consume.
Hand-writing MJCF does not scale to catalogs
Genesis natively consumes MJCF and URDF. Authoring MJCF by hand for tens of thousands of assets is impractical for production RL training at the scale Genesis was built to enable.
Calibrated physics drive sim-to-real
Genesis's differentiable physics is only as good as its inputs. Mass, friction, and inertia within domain-randomization variance let trained policies transfer; uncalibrated values silently break transfer no matter how fast simulation runs.
How Rigyd helps
AI-native infrastructure that automates the hard parts.
Native MJCF output for Genesis
Rigyd emits MJCF (.xml) with calibrated inertials, friction coefficients, collision geometry, and solref parameters, ready for Genesis's native MJCF loader. OpenUSD is produced from the same source for the Genesis USD path.
Physics calibrated for sim-to-real
Mass is calibrated within ±15-20%, friction within ±0.1, both inside the domain-randomization variance Genesis tasks already wrap around. Policies trained on Rigyd assets transfer like policies trained on hand-tuned assets.
Catalog-scale generation
Bulk API processing turns tens of thousands of 3D models, images, or text descriptions into Genesis-compatible assets in hours rather than engineer-months. Diverse training catalogs become a one-step pipeline.
output formats Genesis consumes directly
per asset, from input to Genesis-ready output
mass accuracy, inside DR variance
Feed Genesis the asset diversity its GPU speed unlocks
Drop in 3D models, images, or text descriptions and get Genesis-compatible MJCF or OpenUSD in minutes.
Starts at $29/month. 30 credits included.
Frequently asked questions
Does Genesis natively consume Rigyd's output?
Yes. Genesis consumes MJCF (.xml) and OpenUSD natively, both of which Rigyd emits directly. The MJCF output includes inertials, friction, collision geometry, and contact parameters tuned for Genesis's fast GPU-parallel physics; the OpenUSD path is available when downstream tooling expects USD. No hand authoring is required after Rigyd's export.
How does Rigyd compare to writing MJCF by hand for Genesis?
Manual MJCF authoring takes ~4 engineer-hours per asset (geometry cleanup, collision decomposition, mass and inertia computation, friction assignment, validation). Rigyd compresses that to ~5 minutes per asset with mass within ±15-20% and friction within ±0.1 of measured. Override either when you have lab measurements; for the long tail of catalog assets the AI estimate is inside the domain-randomization band policies already generalize across.
Can Rigyd outputs drive Genesis's differentiable physics?
Yes. Genesis's differentiable physics requires calibrated mass, inertia, and friction to compute meaningful gradients. Rigyd-emitted MJCF carries inertial tensors derived from volume × material density and friction coefficients keyed to surface material, both of which feed straight into Genesis's differentiable solver. Override the values where you have ground truth; let the AI estimate cover the long tail.
Does Rigyd support Genesis's soft-body and fluid features?
Rigyd's primary output today targets rigid-body simulation: rigid mass, friction, restitution, collision meshes, and convex decomposition. Soft-body parameters (Young's modulus, Poisson's ratio) and fluid properties are not estimated automatically and need to be configured by hand on top of the Rigyd geometry. Most Genesis users start with rigid-body workflows where the bottleneck is asset diversity rather than constitutive parameters; Rigyd targets that bottleneck.
Can I use Rigyd in a Genesis training pipeline at scale?
Yes. The Enterprise API processes thousands of source assets in parallel and writes MJCF or OpenUSD outputs to your storage of choice, ready to be referenced by Genesis's asset loader at training time. Generation runs decoupled from training, so a single overnight batch can produce the catalog diversity a multi-day GPU-parallel Genesis training run consumes.
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
Physically 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