Information Fusion Algorithm Of Pose Measurement Based On Double-IMU And Vision

Obtaining the attitude of a measured object quickly and accurately has great significance in aerospace and industry manufacturing fields.IMU has high update frequency,while the errors of IMU diverge with time,therefore it doesn't support long-term measurement.Vision pose measurement's errors don't diverge with time,but its' measure speed depends on the camera's photographing frequency.What's more,it fails when the line of sight is obscured.This paper combines the features of IMU and vision measurement,fast and accurate pose measurement algorithm based on double-IMU and vision is discussed.The main contents and key techniques of this dissertation are as follows:1.The algorithm of double-IMU pose measurement is discussed.While the movement of platform is unknown,one IMU is fixed to the target,the other is fixed to the platform,measuring the angle speed of target and platform.This paper uses Euler angle differential equation to calculate object's pose.Orthogonal double-vector calibration and q-method are used to calibrate the coordinate system.2.Vision pose measurement technology is discussed,which can be divided into two parts to iterate: pose calculation and feature points matching calculation.When the matching relationship between photography feature points and space feature points is unknown,the orthogonal iteration pose measurement algorithm with feature points matching is studied.The algorithm can calculate pose and matching relationship at the same time.3.Information fusion algorithm of pose measurement based on double-IMU and vision is discussed.A multi-speed adaptive extended Kalman filter algorithm is proposed to fuse the measurement results of double-IMU and vision,which consists of gain calculation,measurement update and adaptive fading memory forgetting factor calculation.Experimental results show that the fusion algorithm can measure object's pose relative to a moving platform steadily and accurately.