Niantic filed 3 patents this quarter, all in VR & AR.
The applications focus on building and maintaining accurate 3D maps of the real world to enable precise AR experiences. Technologies covered include automated validation systems that score the reliability of mapped points of interest, server infrastructure that combines crowdsourced scan data from users into unified map tiles, and AI-powered camera localization that determines user position through feature sampling and foundation models.
Niantic's 3 VR & AR patents tackle the challenge of building and validating accurate real-world maps for augmented reality from different angles. One application describes a validation system that cross-checks localization results from independent frame pairs against sensor-derived ground truth, filtering out false positives that single-query confidence metrics would otherwise miss and producing reliability scores for each mapped location. Another addresses the crowdsourcing problem by using control points and gradient descent optimization to align and merge overlapping scan data from different users' devices into coherent 3D map tiles on the server side. The third takes a different approach entirely, using random sampling of visual feature tokens to create map representations with fixed computational cost and pairing them with a foundation model that localizes cameras across any location without requiring expensive 3D reconstruction or location-specific retraining.
All data sourced from USPTO patent filings. Google Patents may take several weeks to index recent publications. If a link is unavailable, search for the patent number at USPTO Patent Public Search.