Ground Truth: Why Visual Positioning Systems Are the Solution When GPS Fails
For the last two decades, the question "Where am I?" has been answered almost exclusively by satellites. Global Positioning Systems (GPS) and Global Navigation Satellite Systems (GNSS) are feats of engineering that have reshaped the global economy.
But they were designed for a world of open skies and approximate locations, not for the precise, obstructed, and autonomous world we operate in today.
In the "ground truth" of daily operation, GPS is often fragile. In urban canyons, signals bounce off skyscrapers, causing your blue dot to drift blocks away from your actual location. Indoors, signals can vanish entirely. In rugged terrain or underground, they are non-existent. And in increasingly common security scenarios, signals can be jammed or spoofed.
Niantic Spatial’s Visual Positioning System (VPS) is designed to solve these challenges. We are moving beyond reliance on a single, fragile signal from space to a robust, multi-sensor approach that works in the most demanding environments including inside buildings and harsh terrain.
What is Niantic Spatial VPS?
Niantic Spatial VPS is an engine designed to generate precise location estimates everywhere, including in GPS-challenged and GPS-denied environments.
Our multi-layered, deep learning-based architecture leverages a variety of 2D and 3D data – collected from the ground, air, and space – to deliver a reliable position even when GPS can't. Onboard sensors and real-time computer vision further enhance positioning, orientation, and tracking.
Niantic Spatial VPS provides the fidelity required for both standard 3DOF localization and full 6DOF pose, ensuring the system can support everything from simple wayfinding to complex augmented reality overlays and robotic navigation.
Multiple Data Sources
Traditional approaches to fixing GPS gaps have historically relied on expensive, dedicated hardware or brittle computer vision techniques that break when the environment changes, or even when things like weather or lighting vary. Niantic Spatial VPS takes a different approach: redundancy and robustness through a flexible deep learning-based architecture backed up by real-time computer vision and on-device sensor fusion.
Our system is powered by our Large Geospatial Model (LGM), a geo-specific model of the world that leverages billions of images to maintain a fresh, reliable view of reality. It fuses multiple inputs:
-
Basemap: Traditional 2D and 3D geospatial datasets, including 2D maps and 3D terrain and surface models.
-
Enhanced: Ground and low altitude scans from mobile and aerial platforms.
-
On device enhancement: Accelerometer, magnetometer, real-time video and LIDAR.
This combination allows the system to serve a wide range of users – from public sector and defense teams operating in contested environments – to logistics companies tracking autonomous delivery bots in dense city centers.
Critically, the system is designed to function hierarchically. In areas where we have enhanced data – the system delivers its highest fidelity, enabling object-level interaction. But operating only with global basemap coverage, the system can deliver reliable locations in GPS-denied areas.
| Feature | Niantic Spatial VPS | Others |
|---|---|---|
| Data Sources | Deep learning based architecture enables localization from a variety of 2D and 3D data sources. | Require pre-mapping using proprietary software or use of closed proprietary datasets. |
| Robustness | Industry-leading research underpins a redundant deep-learning based system that ensures robust performance. | Rely on brittle CV approaches that may not be reliable in varying lighting and weather conditions. |
| Pipeline | Pipeline produces localization maps and 3D digital twins with semantic content. | Pipelines are typically single purpose for localization and produce only limited mesh outputs. |
The Future of Localization
We are currently scaling Niantic Spatial VPS to support the localization of AI agents, robots, and mixed reality headsets virtually anywhere, offering a unified API that offers precise localization in high-value areas and reliable fallback coverage everywhere else.
Achieving true universality requires continuous enhancement of our data foundation, and the Niantic Spatial platform is built to integrate new, high-fidelity data streams from strategic partners over time. This strategy is exemplified by our recently announced partnership with Vantor to deliver an integrated positioning architecture. This solution leverages Vantor's Vivid Terrain global terrain map to provide precise positioning and orientation in Niantic Spatial’s VPS. This integration connects air and ground platforms to a unified, resilient coordinate system, significantly advancing our ability to provide precise positioning and orientation in GPS-challenged environments globally.
This partnership will power critical, new capabilities, supporting precision wayfinding for devices, enabling navigation for autonomous industrial robotics, and providing resilient coordination for field operations.