Spatial Computing
Computing systems that understand and operate within the three-dimensional structure of a physical space. Combines computer vision, sensor fusion and 3D mapping so that software can reason about where things are, not just what they are.
Also known as: 3D computing, mixed-reality computing
Spatial computing is the umbrella for systems that turn a physical space into structured digital data and then operate on it. The output is a persistent 3D map of a room, store, factory or street, annotated with positions of objects, people and devices. Once that map exists, software can answer questions a flat camera feed cannot: where is the customer standing, which aisle is the misplaced product on, can the robot navigate around the spilled coffee without bumping into the shelf.
The hardware stack is broad. Phones with LiDAR sensors, cameras with depth perception, smart glasses, robot-mounted scanners and warehouse fixed-camera arrays all feed raw observations into a spatial pipeline. The processing side runs computer-vision algorithms (structure-from-motion, SLAM, gaussian splatting) to convert observations into a coherent 3D model, then a semantic layer tags the model with what each object is. Apple’s Vision Pro, Meta’s Quest, Niantic’s Lightship and Auki’s posemesh all sit somewhere in this stack.
The decentralised-AI angle matters because spatial data is privacy-sensitive in ways that flat image data is not. A 3D map of a retail store includes employee movement patterns, customer routes, dwell times in front of shelves and inventory positions. A 3D map of a home tells you where every camera and microphone is, where children sleep and where valuables are stored. Centralised spatial platforms aggregate this data; decentralised approaches like the posemesh push the storage and the inference toward the venue owner. Whether the decentralised promise holds depends on operator distribution, key custody and what the protocol’s cloud-AI fallback actually sees.
Spatial computing is also where physical robotics, AR/VR headsets and autonomous machines start to share infrastructure. The same 3D map a warehouse robot uses to navigate is the map a head-mounted device uses to anchor virtual annotations. The Intercognitive Foundation (peaq, Auki, Geodnet, Mawari, Tashi) is the closest thing to a DeAI-native consortium framing this shared infrastructure as a single coordination problem.