◈ Spatial
Spatial is a multi-sensor, multi-camera, multi-projector platform for deterministic spatial computing — designed to run on commodity hardware first (an RGB webcam and a projector) with a contract system that scales up to RGB-D cameras, stereo vision, LiDAR, hand tracking, and projector fleets.
🔍 Overview
Architecture
Sensors → Perception → SpatialWorld → Interaction → Projection- Core: world modeling, geometry, perception, interaction and event handling, exposed through a typed
SpatialWorldAPI with capability-based observations and a replay mechanism. - Devices: replaceable sensor and projector adapters, so new hardware plugs in without touching the core.
- Providers: external service integrations.
- AI: semantic tools and capabilities layered on top of the world model.
- Apps: runtime and UI entry points — a FastAPI/WebSocket backend paired with a React/Vite frontend.
Key Features
- Sensor-agnostic core: the same world model runs whether the input is a webcam or a LiDAR array.
- Deterministic runtime: commissioning procedures and a replay mechanism make physical-hardware behavior reproducible.
- Commodity-first: starts working with hardware most people already have.