基于视觉的飞行器偏航角和位置估计与控制

针对惯性导航系统(inertial navigation system，INS)和全球定位系统(global position system，GPS)导航系统的不足，提出一种基于视觉信息的飞行器相对偏航角和相对位置估计方法。采用金字塔(lucas-kanade，LK)光流算法对地面目标上的2个特定特征点进行位置估计，然后根据摄像机成像原理，用高斯-牛顿迭代法估计出飞行器的相对偏航姿态角和相对位置，并设计 PID的位置控制系统，实现飞行器对地面目标的跟踪。仿真结果表明：该方法相对偏航角和位置的估计精度较高，飞行器对目标跟踪准确。

Vision-based Air-vehicle Yaw & Position Estimation and Control

Due to the disadvantages of the INS and GPS, this paper presents a vision-based relative position and yaw estimation strategy for air-vehicle, then use the result to design a PID control system to track the ground target. This scheme uses Lucas-Kanade pyramid optical flow algorithm to estimate two specific feature points’ position and use the Gauss-Newton iterative method to estimate the air-vehicle’s relative position and yaw angle, according to the camera imaging geometry. Then the PID position control system is designed to track the target. The simulation shows that the estimate of relative yaw and position is accurate and air-vehicle tracking the target accuracy. The simulation results show that the method has high accuracy on yaw and position estimation, and the air-vehicle has good performance on target tracing.