Niantic Spatial has signed a memorandum of understanding with the Singapore Land Authority (SLA) to jointly evaluate the technical feasibility of building Singapore's "first Visual Positioning System," a national-scale layer that would let robots, vehicles, and other machines determine their exact position and orientation by matching a camera feed against detailed 3D map data rather than depending on satellite signals. Under the arrangement, SLA supplies the raw material, its existing high-resolution aerial imagery and nationwide 3D point cloud data, while Niantic Spatial applies its geospatial AI, computer vision, and visual positioning technology built on the same underlying platform that powers its consumer-facing VPS work. What makes this specific approach genuinely notable is the data source: rather than sending teams out to scan Singapore street by street purely for this project, the plan is to test whether mapping data originally created for human planning, land surveying, and government operations can be repurposed directly into positioning infrastructure for machines, a meaningfully cheaper and faster path than building a dedicated VPS map from scratch. Niantic Spatial's CEO framed the goal directly as turning the country's geospatial data into "infrastructure machines can use to navigate safely" without that street-by-street scanning effort, while SLA's CEO pointed to the practical upside of pairing existing national datasets with advanced positioning technology to unlock genuinely new applications.
The stated use cases span a genuinely wide range, autonomous navigation and robotics, urban planning and city operations, emergency response and public safety, and location-based digital experiences, and that breadth is exactly why this pilot matters beyond Singapore's borders. A visual positioning system becomes considerably more accurate than GPS in exactly the environments where GPS struggles most, dense urban canyons, indoor spaces, and anywhere satellite signal gets obstructed, which is precisely where construction sites, large campuses, transit hubs, and real estate developments most need reliable positioning for both autonomous equipment and AR-based wayfinding or inspection tools. For any organization working in real estate, construction, or smart infrastructure, this pilot is worth tracking closely, not because it directly unlocks anything today, the current phase is limited to feasibility testing rather than broader rollout, but because it's testing a genuinely reusable model: if a country's existing land survey and mapping data can double as machine-navigable infrastructure without a separate capture effort, that same approach could meaningfully lower the cost of building precise indoor and outdoor positioning for any large facility or development that already has decent existing mapping data sitting unused.