A ground-based vision system for UAV tracking

It is presented a vision system based on a standard RGB digital camera to track an unmanned aerial vehicle (UAV) during the landing process aboard a ship. The developed vision system is located on the ship's deck and is used to track the UAV during all the landing process. A Ground-based vision system makes it possible to use an UAV with small size and weight, since the UAV will have less computer requirements. The proposed method uses a particle filter (PF) for pose estimation and an unscented Kalman filter (UKF) approach for filtering. This combination provides an adapted filtering framework for tracking. The implemented particle filter is inspired in the evolution strategies present in genetic algorithms with modifications in the mutation and crossover operators to avoid sample impoverishment. Results show that position and angular estimates precision is compatible with the automatic landing system requirements.