In this thesis, two of the main technical limitations for a massive deployment of Unmanned Aerial Vehicle (UAV) have been considered.

The Aerial Refueling problem is analyzed in the first section. A solution based on the integration of ‘conventional’ GPS/INS and Machine Vision sensor is proposed with the purpose of measuring the relative distance between a refueling tanker and UAV. In this effort, comparisons between Point Matching (PM) algorithms and Pose Estimation (PE) algorithms have been developed in order to improve the performance of the Machine Vision sensor. A method of integration based on Extended Kalman Filter (EKF) between GPS/INS and Machine Vision system is also developed with the goal of reducing the tracking error in the ‘pre-contact’ to contact and refueling phases.

In the second section of the thesis the issue of Collision Identification (CI) is addressed. A proposed solution consists on the use of Optical Flow (OF) algorithms for the detection of possible collisions in the range of vision of a single camera. The effort includes a study of the performance of different Optical Flow algorithms in different scenarios as well as a method to compute the ideal optical flow with the aim of evaluating the algorithms. An analysis on the suitability for a future real time implementation is also performed for all the analyzed algorithms.

Results of the tests show that the Machine Vision technology can be used to improve the performance in the Aerial Refueling problem. In the Collision Identification problem, the Machine Vision has to be integrated with standard sensors in order to be used inside the Flight Control System.