Kalman Filter-Based Position Estimation in UAV Swarms: Impact of Swarm Size, Flight Paths and Onboard Sensors

This paper investigates swarm navigation performance for unmanned aerial vehicles (UAVs) under varying flight path geometries and swarm sizes. Employing both local and master Kalman filter (KF) estimation techniques, the study analyzes the accuracy and efficiency of each approach in estimating UAV positions. Simulations are conducted across a range of flight paths, from simple linear trajectories to more complex maneuvers, with swarm sizes ranging from small formations to larger groups, and with different sensors such as onboard cameras to measure attitude. The primary performance metric is position estimation error. A key focus of this study is understanding how attitude estimation, derived from an onboard camera mounted on the parent UAV, the quality of the parent UAV’s gyroscope, and the number of child UAVs of the swarm influence both local and master KF performances. Results show that increasing the swarm size does not make position estimations better. However, adding attitude measurements and using a high quality gyroscope on the parent UAV significantly improves position accuracy. This research contributes to the development of robust swarm navigation algorithms for UAVs for applications such as coordinated surveillance, search and rescue operations, and collaborative payload delivery.