A VISUAL-INERTIAL NAVIGATION ALGORITHM FOR MICRO AERIAL VEHICLES USING THE UNSCENTED KALMAN FILTER
João Paulo de Almeida Barbosa1; Raphael Ballet1; Davi Antônio dos Santos1; Stiven Schwanz Dias2
1 Instituto Tecnológico de Aeronáutica - ITA; 2 Embraer
Resumo
This work proposes a position, velocity and heading estimation algorithm based on the Unscented Kalman Filter (UKF) for a multirotor Unmanned Aerial Vehicle (UAV) using low-cost components, such as a gyro-stabilized two-axis gimbal platform containing a downward-facing onboard camera, inertial sensors and an ultrasonic range sensor. For addressing the navigation estimation problem, the inertial sensors were used in the prediction equations of the UKF while measurements from image processing and ultrasonic sensor readings were applied in the update step of the filter. This method is evaluated via extensive Monte Carlo simulations, which show its effectiveness. Furthermore, this paper shows a performance comparison between the proposed UKF based method and an equivalent EKF based method. The proposed method shows a slightly better performance, but has a higher computational burden.
Palavras-chave: Unscented Kalman filter; visual-inertial navigation; sensor fusion; unmanned aerial vehicle