Autonomous Flight Control Of A Quadrotor UAV Based On Visual Servoing

The quadrotor UAV has many advantages,such as small size and vertical take-off and landing.It have various flight mode such as fixed-position hovering and target tracking in complex environments,and is widely engaged in aerial photography,search and rescue,patrol and other tasks.Thanks to the robot technology,the emergence of high performance sensors,new materials,embedded devices and other technologies has also accelerated the development of the quadrotor UAV industry.Researchers improve the performance of quadrotor by developing advanced control methods and appropriate sensors,which can replace the manual operation to achieve autonomous flight.This paper mainly studies the autonomous flight of quadrotor UAV based on image-based visual servoing.The quadrotor has dynamic characteristics of underdrive and nonlinearity,which presents a challenge to image based visual servoing.At the same time,the uncertainty of image feature depth and external disturbance brought by monocular camera is also the problem to be faced.Aiming at these problems,the control method based on visual servoing in quadrotor UAV with airborne camera as main sensor is studied.The main research content and contributions can be summarized as the following three points:This paper defines the virtual camera coordinate and generates the virtual image plane.The image-based visual servoing control will be processed in this plane.The virtual image plane is always parallel to the horizontal ground so that the ground target has the same depth value.By perspective projection,the appropriate image moments are selected on the virtual image plane as the image features.The kinematic model of image features is derived for autonomous flight control of the quadrotor UAV based on visual servoing method in the later.Aiming at the fixed-point hovering of quadrotor UAV based on visual servoing,the finite-time controller is designed.A finite-time control method based on non-singular fast terminal sliding mode is proposed.The inner-loop attitude and outer-loop position finite time controller of quadrotor based on visual servoing are designed respectively.In addition,the uncertainties of images and external disturbances are considered.The compensation terms are introduced into the controller to deal with these uncertainties.The finite-time convergence of the system is proved by Lyapunov analysis,and the image feature error finally converges to zero.Aiming at the moving target tracking based on visual servoing,an image-based moving target tracking controller is designed.The proposed controller updates the time varying depth information of the camera and the linear speed information of the moving target through the adaptive law,which can realize the moving target tracking the unknown ground speed.