Image-based Visual Serving Of A Quadrotor UAV

Because of their unique structure,quadrotor UAVs are relatively small and low weight.They can takeoff and landing in narrow space and operate agilely in cluttered and complex environments.They have various flight mode such as hovering,three-dimensional flight,and even operate close to the ground.They can be wildly used in shooting aerial photos,search and rescue,topographic survey,military reconnaissance.Thanks to the rapid development of high performance hardware devices,embedded systems,sensors,new materials and other related technology,highly intelligent and autonomous robot systems became possible.When equipped with appropriate sensors and algorithms,quadrotor can function autonomously,and be deployed for complex and dangerous tasks.Thus,there is a great research interest in autonomous quadrotor unmanned aerial vehicles.This paper studies the Image-based visual servoing(IBVS)of a quadrotor UAV.IBVS method uses visual feedback from the onboard camera to control the quadrotor motion in real-time.The quadrotor is equipped with a monocular camera and a standard inertial measurement unit(IMU).The main research content and contributions can be summarized as the following three points:1.An observer for the translational velocity of the quadrotor and an IBVS controller based on backsetpping are designed.Because the translational velocity of the quadrotor cannot directly obtained from measurement,we designed a velocity observer using visual feedback form the onboard camera.The observer is proven to be exponential stable.Then,a nonlinear controller based on backsetpping technique is proposed.The error of observer is considered in the controller,and the stability of the controller is analyzed using Lyapunov analysis.The IBVS controller can be applied to inclined target plane case with only three point features,thus makes it useable in more general environment.2.A method that directly planning and tracking in image space is proposed to solve the IBVS of a quadrotor UAV.By elaborately analyzing the relation between the Euclidean space and the image space,the one to one mapping between the image features and the quadrotor pose is found,which can gracefully transform the visual servoing problem to a feature tracking problem,and consequently improve its performance.Then,a trajectory tracking controller is proposed,and the stability of the controller is analyzed using Lyapunov analysis.To formulize the visual servoing problem as an image feature tracking problem,this method held intrinsic advantage over conventional position control method and improved visibility was achieved.3.An Image-based visual tracking(IBVT)controller is proposed for the moving target tracking problem of a quadrotor.Consider that the target is moving with unknown but bounded linear velocity,linear acceleration,yaw velocity,and yaw acceleration,an IBVT controller is designed and proven to be uniformly ultimately bounded(UUB).The proposed controller does not need the time varying depth information of the target with respect to the camera.In addition,the linear velocity of the target are not needed.Finally,to test the validity of the proposed controllers,various simulations are performed.The proposed IBVS controller are also tested using a quadrotor UAV platform.Experimental results show that the proposed methods achieve the IBVS control of a quadrotor successfully.The proposed methods will be beneficial in applications such as monitoring and surveillance.