Nianticspatial

The product is a set of real-world foundation models for physical AI that reconstruct geometrically accurate digital twins, provide vision-based positioning where GPS fails, and enable semantic querying and reasoning at every 3D point. It sells to developers, engineering and data teams building robotics, autonomous systems, and field operations in industries such as robotics, defense and intelligence, and oil and gas, including enterprises and organizations operating beyond screens. It is delivered as infrastructure APIs and SDKs with capture tools for ground and aerial data.

Key features

• Create high-fidelity digital twins as meshes
• Output Gaussian splats with SPZ format
• Vision-based positioning without GPS
• Capture 3D data from ground and aerial sources
• Query world through semantics at every 3D point
• Measure and reason about 3D points
• SDK and APIs for developers
• Scaniverse capture application