Visual Servo-based Grasping Control Of Aerial Manipulator

In recent years,the technology of rotor UAV has been more and more mature.For its simple operation,flexible and stable features,UAV have been applied successfully in many fields,such as agriculture,military,disaster relief and so on.However,most applications of UAV focus on observation by aerial photo l acking interaction with environment.If a multi-rotor UAV is equipped with a flexible robot arm,it can grasp and transport objects in the external environment.The existing UAV grasping researches either require external device such as the motion capture system with high cost but limited coverage to obtain real-time location information of UAV and target,or just implement slow flight and grasping based on visual,which is unsuitable for large-scale application.Therefore,the main task of this dissertation is to realize the high-speed autonomous grasping of UAV without motion capture system.In this dissertation,the UAV is equipped with a rotatable camera and a lightweight robotic arm.A complete three-dimensional dynamic model of the UAV-arm composite system is established in the Cartesian coordinate system by using Euler-Lagrange equation.Without loss of generality,the dynamic model is simplified to the x-z plane.Furthermore,the manipulator is simplified to single degree of freedom.Based on the geometric analysis of the UAV-camera system,mapping between target image position and the UAV space position are established in the x-z plane.The attitude of the UAV and camera is decoupled with the target position in the image by projecting the image onto the calibrated virtual horizontal image plane.The dynamic model of the x-z plane is mapped into the image space by the geometric relation and the differential smoothness of the system is proved in image space.Based on the image space mapping model,the dissertation proposes an image space trajectory planner separately utilizing the minimized Snap and the iterative linear quadratic regulator(i LQR)algorithm,which are able to transform the various coupling constraints in the flight process into an optimization problem.In order to track the desired trajectory,a visual servo controller is designed with the image pixel errors of ta rget selected as the feedback.An inner and outer loop PID controller is also designed as a supplement for tracking.To verify the proposed algorithm,a simulation platform was built for UAV-arm system based on Gazebo.The effectiveness and robustness of the model,trajectory planner and tracking controller were validated on the platform.This dissertation explores a new research direction on UAV grasping,and realizes the trajectory planning and tracking control of UAV grasping based on visual servo.It provides important reference for other researchers in the UAV grasping field and has a great significance to application research of China's UAV transportation.