Research And Application Of Autonomous Landing System For Mini-Quadrotor Based On Visual Feedback

The mini-quadrotor involves various aspects such as navigation,computer vision,control theory,and embedded electronics,which has the characteristics of small size,low cost and flexibility.It has been widely applied in agriculture,military,monitoring and other fields.It is an ideal theoretical and applied research platform.The pose estimation is one of the key factors for the operation and control of mini-quadrotor.The positioning method based on GPS is widely used in the mini-quadrotor positioning system.However,in some specail situations,such as indoors or in buildings,the positioning accuracy of mini-quadrotor is low or failure because of the unstable signal of GPS navigation system.Vision assisted positioning has the advantages of simple structure,stabe signal,light weight and low price.This paper studies the autonomous langding syetem of mini-quadrotor based on visual feedback.It has certain theoretical and practical significance.The research of this paper is as follows:Firstly,aiming at an EKF for attitude estimation algorithm based on conjugate is designed for the problem of low accurary of attitude estimation based on Extended Kalman Filter.In this paper,the conjugate gradient algorithm is introduced into EKF to reduce error caused by linearization of nonlinear systems.Based on that,the non-gravity motion acceleration is introduced into the noise covariance matrix,which realizes self-adaptive and dynamic estimation of noise.And the yaw angle is calculated by the method of inclination compensation.It improves accuracy and anti-jamming ability.Then,aiming at the problems of limited vision assisted landing accuracy and application limitations,a target recognition algorithm based on HSV color model and Hough transform is designed.And a pose estimation algorithm based on visual feedback.A simple and easy to recognize cooperative icon is designed and using segmented icon size,a target recognition algorithm based on HSV color model and Hough transform is implemented.Based on that,according to the projection model of camera and the geometrical characteristics of the icon.A pose estimation algorithm based on visual feedback is realized,which effectively improves the range of use of the icon,the recognition rate,and the accuracy of pose estimation.Finally,a mini-quadrotor automous langding experimental platform based visual feedback is built to verify the algorithm and the designed system in this paper.The results show that the improved EFK algorithm proposed in this paper has higher precision and stronger anti-jamming ability than the standard EKF algorithm.The range of cooperative icon application,the target recognition rate and the accuracy of the pose estimation algorithm based on visual feedback is improved.It meets the requirements of the visual landing of the mini-quadrotor.