A Robust Adaptive Filter Estimation Algorithm for Vision-Based Cooperative Motions of Unmanned Aerial Vehicle

The thesis intends to propose a robust adaptive time-varying Kalman Filter (KF) estimation algorithm for a class of Multi-input Multi-output (MIMO) uncetain system. The proposed algorithm combines a time-varying KF with an adaptive neural network. The KF is used for estimating the status of target, and the adaptive neural network overcomes the uncertain factors, being trained by two error signals for the purpose of improving the robust performance of this proposed algorithm, and the bounded of the estimation error is proved by Lyapunov theory. Finally, the state estimation of lead aircraft (leader) in the unmanned aerial vehicle (UAV) cooperative motions is designed on the basis of the proposed method. Simulation test demonstrates that this proposed algorithm can estimate the state of leader which contains uncertain factors, and the wing aircraft (follower) can cooperate with the leader well, then the effectiveness of this algorithm is validated.