四轴飞行器视觉组合导航系统设计与实现

Design and Implementation of Visual Navigation System of Quadrotor

为实现四轴飞行器的自主飞行,设计了该视觉导航系统;采用基于ARM处理器的飞行控制器和导航控制器的双CPU结构,提高了系统的运行速度;飞行控制部分采用四元数解算姿态,运用经典的PID控制设计了X、Y、Z三个轴的PID控制器进行整个系统的飞行控制;导航部分采用不敏卡尔曼滤波(UKF)融合惯导位置和视觉位置,从而给出载体最优位置,提高导航精度;实验结果表明,基于图像和惯性导航的视觉组合导航方式可使导航精度保持在0.5m内,同时整个系统具有较好的快速性和稳定性。

In order to realize autonomous flight of quadrotor,a vision navigation system was designed.Double CPU structure was adopted,which combined flight controller with navigation controller based on ARM processor and improved the speed of the system.Flight control part adopted quaternion algorithm of attitude,using the classic PID control design of the X,Y,Z three-axis PID controller for flight control of the system.Navigation part used unscented Kalman filter（UKF）to fuse inertial position and visual position,which gave the optimal position of the carrier and improved the navigation precision.Experimental results show that the navigation accuracy can be kept within 0.5m,while the whole system has better speed and stability based on images and visual navigation inertial navigation mode.