Monocular Vision Based Visual Servo Control Research Of Quadrotor UAV For Tracking The Ground Target

Quadrotor unmanned aerial vehicle(UAV)has attracted wide attention in military or civilian fields recently due to its rapid manoeuvrability,precise hovering ability.At present,it is being developed for the implementation of target reconnaissance,precision strike,infrastructure monitoring and other important tasks.In which,target tracking is a vital part for the autonomous execution of these tasks.During the design process object tracking arithmetic,camera is the main sensor for collectting target information.This paper aims to realize real time tracking of ground target of quadrotor platform using only camera and inertial navigation system.In order to avoid the influence of camera calibration error and the recognition difficulties of prior target geometric model,as well improve the practicability of the algorithm,this paper designs a tracking algorithm of quadrotors for ground targets under the image based visual servo frame.For the image singular of traditional image features in servo control,the perspective projection image moments of the target image are firstly extracted as the image feature.And the virtual camera is introduced to isolate the uncontrollable roll and pitch motion and ensure the stability of the quadrotor three-dimensional(3D)motion.Via analyzing the selected image feature motion and quadrotor motion,an integrated image-quadrotor model is established in the presence of multi-source interferences.Using the established model,considering the affect of multi-source interferences and the object lost caused by violent attitude motions in the process of servo tracking of quadrotor,we design the attitude constraint anti-distubance visual servo control strategy to stabilize the quadrotor motion and ensure the target holding in the field of view of carmra.As for the effect of high dynamic and multi-source interferences,a robust integral of the sign of the error(RISE)controller is constructed for the servo loop(outer loop)and attitude loop(inner loop)of quadtotor respectively,to improve the anti-interference ability of the system and ensure the stable motion of quadrotor.For each loop,the controller includes a model feedforward control term to compensate for the system model nonlinearities,a linear feedback control law to stabilize the system,and a robust nonlinear integral term to suppress multi-source disturbances.In addition,considering the difficulty of linear velocity measurement,a linear velocity-free visual servo control is constructed based on the expected compensation technology.When the target enters the stationary phase,the fourrotor is required to precisely servo to the top of target for executting complex tasks.In this phase,considering the target loss problem caused by the sharp attitude motion,we design the strict attitude constraint control method by a fusion of the saturation command filter based attitude command constraints and the integral barrier Lyapunov function(iBLF)based attitude response limits,simultaneously guaranteeing the roll and pitch angles stay within the preset state space all the time,and realizing strict attitude restrictions.