Flying without GNSS: LiDAR SLAM and visual-inertial odometry
When satellite navigation is gone, the aircraft has to work out where it is from what it can sense. This is how.
Inertial sensors alone drift
Accelerometers and gyroscopes measure motion, and integrating them gives position. Small errors grow with time, so an inertial-only solution slowly drifts away from the truth. Triple-redundant IMUs reduce the risk of a faulty sensor, but they cannot stop drift on their own.
Visual-inertial odometry
Cameras track features in the scene from frame to frame. Combined with inertial data, this gives a motion estimate that is much more stable over time. It works best with texture and light, which is why it pairs well with a second method.
LiDAR SLAM
LiDAR measures distances to the surroundings and builds a map while locating the platform inside it. It works in darkness and in low-texture scenes, and it is especially strong on ground vehicles and low-flying aircraft.
Why FPGA acceleration matters
Both methods are compute heavy. Running feature tracking, point cloud matching and fusion in FPGA hardware keeps them in real time without taking time away from the control loop. That is how Speculis keeps navigating when GNSS is denied.
PRODUCT RENDER