GNSS-Denied Localization

The field studies how robots, vehicles, aircraft, and people can estimate their positions when satellite navigation is blocked, degraded, or absent. Common approaches combine inertial dead reckoning with visual-inertial or LiDAR SLAM, using Kalman filters and factor graphs to limit drift; absolute cues from maps, landmarks, radio ranging, magnetic fields, or motion constraints can further correct position. Learned visual matching and sensor-fusion models are increasingly paired with geometric and probabilistic estimation for settings from urban streets and tunnels to forests, underwater missions, and UAV flights. A central challenge is maintaining accurate, real-time localization as sensors or maps degrade and environments change, while validating performance beyond controlled datasets and across devices and deployment conditions.