Online Control For Two-axis Pan-tilt-Unit Of Micro Aerial Vehicles

In recent years,with the advancement of unmanned aerial vehicles’(UAV)technology,the swarming,miniaturization and low cost of UAVs have become an important development trend.When UAV swarm perform tasks,such as anti-terrorism reconnaissance,precision strikes,etc.,the single UAV in swarm should have the ability to perception environments with the electro-optical PTU.However,low cost micro flying vehicle’s computation ability and communication bandwidth are limited.The traditional PTU control method which relies on links and sensor operator is no longer applicable.In order to reduce the operational burden and communication burden in the PTU control process,improve the autonomy of the UAV,and enhance the mission capability of the UAV in the communication rejection environment,this thesis conducts an in-depth study on the online control technology of the PTU,including the feedback for target states,the control law for PTU attitude,and the experiment for validating this thesis’ s methods.The main work and achievements are as follows:(1)A fast target recognition method under onboard limited resources is proposed.Aiming at the requirement of real-time target detection under finite airborne computing resources,this thesis firstly uses spectral residual saliency to extract significant regions,and then uses shallow neural network to classify and identify targets.This thesis proposes a two-scale method that can effectively adapt to the detection and identification of different scale targets.In addition,the information entropy is used to eliminate the image that does not contain the target,which further improves the real-time and accuracy of the target detection and recognition.Finally,the characteristics of the target state feedback are analyzed as well.(2)The PTU gaze tracking control law in high dynamic scenes is designed.Firstly,by analyzing the relative pose relationship between the UAV,the PTU and the target,a kinematic model of the target image plane coordinates is established.Then,three control laws are designed based on the target image plane error,the PTU phase plane error and the line-of-sight point error respectively,Furthermore their control performance is analyzed by numerical approaches.(3)The high-fidelity semi-physical simulation environment based on Airsim and physical system are constructed,and the proposed online control method of the PTU is verified by experiments.The experimental results prove that the target detection and recognition method proposed in this thesis has practically feasible real-time and accuracy for real-world scenarios.The control method based on the target image plane error and the PTU phase plane error has an advantage in the unbiasedness of the control,and the control method based on the line of sight point error can maintain the robust tracking of the target during the rapid change of the UAV attitude.