Design Of Vision Based Robot Flying Manipulator System

At present,the multi-rotor unmanned aerial vehicle(UAV)are mainly used in aerial mapping and plant protection,etc..With the development of UAV technology,the research of flying robot with operational capability have become one of the research focal.The robot flying manipulator system(RFMS)can perform the tasks of aerial capturing,which expands the application fields of conventional UAV.The thesis focuses on the positioning and control problems of visual-guided RFMS when performing grabbing operation in the air.Firstly,the hardware system of RFMS is constructed,while the kinematics and dynamics models of the corresponding modules are analyzed.The RFMS is composed of a flying robot,a working device and visual sensors.The working device is a three degree of freedom(DOF)manipulator.In order to perform grabbing operation,the manipulator is mounted on the bottom of the flying robot.On one hand,one camera is used to self-positioning for the flying robot,on the other hand,the Eye-in-Hand mode is used by adding another camera at the end of the manipulator to achieve the positioning of the target.Furthermore,the dynamic models of the flying robot and manipulator are established using Newton-Euler equation,and the forward and inverse kinematics equation of the manipulator are analyzed.Secondly,an attitude and position estimation algorithm based on extended kalman filter-complementary filter(EKF-CPF)is designed.The RFMS is susceptible to disturbance,which can cause the divergence of EKF.The CPF is used to monitor the attitude calculation of the EKF.Considering the distortion of translational velocity estimation in Pyramid Lucas-Kanade(LK)optical flow algorithm can be caused due to the motion around the yaw axis,a rotation compensation algorithm is designed to correct this problem.In the process of operation,the fast Discriminative Scale Space Tracking(fDSST)algorithm is used to estimate the position and scale information of target.Thirdly,for the disturbance caused by the manipulator motion to flying robot,a novel stratified motion control algorithm with disturbance compensation is proposed.The controller can be divided into two layers,In the top layer,the disturbance of the manipulator motion is described as an external force disturbance to the flying robot,moreover,the disturbance is estimated by Newton-Euler iteration method,and the compensation of attitude angle is calculated according to the dynamics of the flying robot.In the bottom layer,the controller can track the compensated desired attitude angle by decoupling attitude control,while suppressing the disturbance of position and attitude caused by manipulator motion.Based on the work mentioned above,the grasping strategy of RFMS is designed,a search method is designed to re-track the lost target.Finally,the performances of the RFMS is verified in the indoor and outdoor environment.In the indoor environment,OptiTrack is used to verify the accuracy of fixed-point hovering and the effect of disturbance suppression algorithm.The quantitative analysis results show that the algorithm can estimate the compensation amount of the attitude angle and improve the trajectory tracking ability of the flying robot,effectively.In the outdoor environment,the grabbing experiment is carried out,and the results show that the stratified motion control algorithm can suppress the disturbance.Furthermore,the task of capturing target is achieved in the air.