基于计算机视觉的UAV自主空中加油半物理仿真

设计并实现了一种基于计算机视觉的无人机（UAV，Unmanned Aerial Vehicle）自主空中加油仿真平台，采用点匹配与位姿估计算法对软管式加油锥发光二极管（LED，Light Emitting Diodes）标志进行识别.对大气扰动和尾涡流进行了建模，建立了加油锥模型，并设计了加油锥运动状态预估算法.在此基础上，设计了基于线性二次型调节器（LQR，Linear Quadratic Regulator）的飞行控制律.基于上述技术，在实验室环境下设计开发了基于单目机载摄像机的无人机自主空中加油半物理仿真平台.半物理试验结果表明：所设计的位姿匹配算法和飞行控制律能够满足复杂环境下无人机自主空中加油的性能要求，具有较强的实时性、准确性和鲁棒性.

Hardware-in-loop simulation platform for UAV autonomous aerial refueling based on computer vision

A hardware-in-loop simulation platform of unmanned aerial vehicle(UAV) autonomous aerial refueling based on computer vision was developed. Point feature matching and LHM pose estimation algorithms were adopted to the recognition of light emitting diodes(LEDs) on the drogue. The models of atmospheric disturbances and wake vortex had been built, and the drogue model was also presented. Subsequently, a linear quadratic regulator (LQR) control law for UAV was designed. The hardware-in-loop simulation platform with airborne monocular camera for UAV autonomous aerial refueling had been developed in lab environment. Series of hardware-in-loop experimental results demonstrate the feasibility and effectiveness of our presented approaches in this simulation platform.