Vision-Based Dynamic Planning And Control Of Aerial Manipulator

With the rapid development of UAV technology and the maturity of related fields,for example,aerial photography,power inspection military investigation,disaster site survey,large-scale environmental modeling and etc.,UAVs are receiving more and more public attention and gaining strong developmental momentum.However,they are all confined within the domain of environmental monitoring.Therefore,more and more researchers devote themselves to this field and try to endow UAVs the ability to exert an active influence on the environment,tasks such as high-altitude debris removal,shipping of goods,and so on.One of the hotspots of these researches,which called aerial manipulator system,is to mount a multi-degree-of-freedom manipulator on the body of the UAV which can perform more complex tasks through the operation of the manipulator,However,because of the introduction of a manipulator,the coordinated control of such a complex system results from the coupling effect of the UAV and the manipulator press for a solution,which further increases the difficulty of this research.This dissertation focuses on the grasping control of an aerial manipulator system and the ultimate purpose is to achieve accurate,quick and hovering-needless of a target in actual circumstances.To solve the objectives mentioned above,this dissertation proposes a 3-DOF manipulator that is flexible and light in weight,and it moves the structural mass center up to the base of the manipulator through the timing belts,so that it features a more stable center of gravity and structurally limits the influence imposed by the movement of the manipulator on the UAV.As for the tracing of target,a monocular camera driven by servo motor is mounted in front of the UAV.The UVA together with the camera are regarded as a 2-DOF virtual manipulator which is called UAV-camera virtual manipulator in this dissertation.The author also designs a position-based visual servo motor actuator for this UAV-camera virtual manipulator which can keep the target at the center of the camera image all the time through the relative positioning control of the camera.During the whole grasping process,the trajectory planning of the UAV and the manipulator are carried out respectively in Cartesian space.However,in real circumstances,the trajectory tracking of manipulator can be affected by the motion of the UVA,a feedforward control method based on kinematics compensation is proposed to ensure the accurate grasping of the target in motion.All the methods mentioned above have been verified on a physical aerial manipulator platform and have shown good performance to the manipulation.This dissertation focuses on the study of the key issues occurred during the grasping process of the UAV-camera virtual manipulator,so as to achieve the accurate,quick and hovering-needless of a target in actual circumstances,which endows a great referential significance on actual grasping tasks of aerial manipulator system.