光学辅助无人飞行器近进回收属无源工作模式，具有不易受干扰、分辨率高、成本低等独特优势，成为重要研究对象。针对飞行器视觉近进回收的最优路径受动基座、大气、光学等不确定、非线性因素扰动，严重威胁回收的自主性与安全性，本课题开展以下3方面的研究：1）研究球面空间最优交互多模型视觉近进回收测量理论与自抗扰模型，通过观测回收过程的不确定、非线性扰动量，构建非线性视觉-惯姿近进回收抗扰模型，设计多模型融合的误差最优约束条件，以快速调整飞行器姿态进入最优回收路径；2）研究近进回收装置最优交互多模型的长时距跟踪测量与姿态反演，通过检测回收装置全姿态时空显著性特征，观测姿态运动参数并进行半监督P-N闭环分类、学习，优化抗扰回收模型，并提升姿态修正数据精度、连续性；3）半实物仿真验证与外场试验，验证视觉近进回收的关键技术、相关理论和算法可靠性。本课题将为无人飞行器在复杂自然条件下自主安全回收提供新方法新手段。

Optical aids unmanned aircraft recovery working in passive mode, with its unique advantages of less susceptible to interference, high resolution, low cost and etc., has become an important field of research. For optimal path of aircraft recovery using vision is affected by moving base, atmospheric, optics and other uncertainties, nonlinear disturbance and other factors, which are serious threats to the aircraft recovery of autonomy and security. The research including three aspects: 1) Optimal state transition matrix interactive multiple (OSTM-IMM) visual measurement theory and auto disturbance rejection (ADRC) visual model research. By observing the uncertain and nonlinear perturbation using visual, building nonlinear visual-inertial attitude unmanned aircraft recovery model, design optimal error constraints to the fusion IMM model, for adjusting aircraft attitude into the optimal recovery path. 2) Recovery equipment measurement and long-time tracking, with aircraft attitude inversion. By detecting significant features of recovery equipment, observing the attitude motion parameters, which will be classified and learned by semi-supervised PN closed loop, optimized ADRC recovery model, and improve aircraft posture correction data accuracy, continuity. 3) Semi-physical simulation and field experiments. The key technology, reliability theory and algorithms verified. This topic will provide new methods and tools for reducing the error uncertainty while UAVs recycling.