GEOFF: Geo-location from Edges, Objects, Foundational data, and a Filter

Ground vehicles with navigation capability (e.g., GPS) can index into foundation data (e.g., Google Maps) to gain situational awareness abouttheir surroundings. When GPS and RF navigation sources are degraded, maintaining situational awareness requires an alternative navigationsource. One alternative source is the foundation data itself. The data contain objects at known 3D locations, which project into cameraimagery at measurable 2D locations depending on the pose (position and orientation) of the camera. We propose to determine the camerapose by matching 2D projections of 3D landmarks by combining two image-based navigation approaches: visual-inertial odometry andlandmark matching. The approach operates on a list of 3D building edges and object locations extracted from foundation data. The odometerestimates the cameras pose, which the landmark matcher uses to predict which edges and objects appear in imagery, and where. The matchersearches tight search windows for specific edges and objects, minimizing computation (allowing time to search for more objects) andminimizing opportunity for false matches (allowing lower recognition thresholds to detect noisier objects). The odometer uses the 2D/3Dmatches to form absolute updates to cancel navigation drift. The result is accurate, non-drifting, GPS-denied navigation and more matches tofoundation data objects.