BIFROST
STARDUST · SPACE

Lunar & Martian Data for Verification & Validation

Enabling space robotics teams to build hazard avoidance systems, validate lunar landing systems and build state-of-the-art navigation systems for autonomous platforms.

Working with the organizations operating in the hardest environment there is

NASA

Pixel-Perfect Labels & World Metadata

Move beyond basic 2D boxes with precise segmentation and rich metadata, helping your AI gain deeper scenario understanding, with no manual labeling required.

TERRAIN CLASSES
Bedrock, sand, rocks and hazards, segmented per pixel.
POSE & DEPTH
Camera pose and per-pixel depth on every frame.
WORLD METADATA
Sun elevation and illumination logged per frame.
Lunar surface simulation

Accelerate Autonomy Development by Unlocking Unlimited Sensor Data

Identify Where Your Perception Systems Will Fail

Generate thousands of variations across lighting conditions, terrain types, and rock distributions to precisely identify where your perception system will fail ahead of physical deployments.

Craft Customized 3D Terrains

Upload custom Digital Elevation Models (DEM), select surface materials, and customize crater and rock distributions to create specific terrains and scenarios.

Lunar Lighting Simulation

Simulate lunar-specific lighting conditions with full control over sun elevation angles, artificial illumination conditions, and specific lunar lighting artifacts such as the opposition effect.

Create Time Series Scenarios

Generate and customize camera traversal paths through the 3D environment to create time series datasets.

Generate Data at 24FPS

Data is capable of being generated in real time using off-the-shelf GPU solutions.

Hazard Avoidance and Landing

Validate descent and landing systems against rock fields, slopes and craters that no probe has imaged yet.

SENSORS
RGB / EOGrayscaleDepthSegmentation
THE VOCABULARY WE WORK IN
DEMterrain relative navigationhazard detection and avoidanceopposition effectsun elevationrover autonomylunar landingMarstraverseverification and validation
FAQ

How do you train perception for a mission that has not flown?

Render the surface or orbit in simulation with pixel and pose ground truth, and train against scenes that have no real equivalent yet.

Can I use my own terrain?

Yes. Upload a Digital Elevation Model, choose surface materials, and set crater and rock distributions to recreate a specific site.

Why is lunar lighting hard for perception?

There is no atmosphere to soften it, so you get hard shadows, extreme dynamic range and effects like the opposition surge. Stardust gives you control over sun elevation and these artifacts so models do not break in operation.

Unlock Unlimited Off-world Datasets

See Stardust generate a dataset for your landing site, rover or orbit.

Book a demo HOW STARDUST WORKS →