Improvement of Shipboard Landing Performance of Shipborne UAV Using Multi-sensor Fusion

Abstract

This paper propose a multi-sensor fusion positioning algorithm based on inertial navigation, gps, monocular vision and Ultra wide band (uwb) to ensure that the uav accurately and stably land on the target point in the near ship region. Establish the relative coordinate system relative position relationship between uav and ship. Converts the measurement data of various sensors into this coordinate system, and establishes the multi - sensor data fusion equation based on Joint Kalman Filter. It also performs the data validation test based on chi-square test of Mahalanobis Distance. Finally, we use AirSim 3d simulation system to simulate the algorithm. Experimental results show that the modified algorithm can calculate the uav's flight position in the near ship segment in real time, and reduce the impact of the increase of measurement error by a few sensors. It prevents the failure of fusion results, and has certain adaptability and robustness.