Visual-Inertial Odometry Using High Flying Altitude Drone Datasets

Positioning of unoccupied aerial systems (UAS, drones) is predominantly based on Global Navigation Satellite Systems (GNSS). Due to potential signal disruptions, redundant positioning are needed for reliable operation. The objective this study was implement and assess a system high flying altitude drone operation visual-inertial odometry (VIO). A new sensor suite with stereo cameras an inertial measurement unit (IMU) developed, state-of-the-art VIO algorithm, VINS-Fusion, used localisation. Empirical testing the carried out at altitudes 40–100 m, which cover common flight range outdoor operations. performance various implementations studied, including stereo-visual-odometry (stereo-VO), monocular-visual-inertial-odometry (mono-VIO) stereo-visual-inertial-odometry (stereo-VIO). stereo-VIO provided best results; 40–60 m most optimal baseline 30 cm. accuracy 2.186 800 m-long trajectory. stereo-VO degraded increasing due degrading base-to-height ratio. mono-VIO acceptable results, although it did not reach level stereo-VIO. This work presented hardware research results localisation algorithms drones that great importance since use autonomous beyond visual line-of-sight will require solutions compensate disruptions in GNSS positioning. data collected published analysis further studies.