Research On Vision-based Autonomous Landing Control On Moving Target Of Quadrotor

Quadrotor UAVs have played an important role in surveying,transportation and other tasks in recent years.Quadrotors are often required to land on various moving targets in applications.The quadrotor would be used much more with the capability of autonomous landing on moving targets.Some research has been achieved in regions of vision servo or communication based ground-air cooperation landing.But it is still far from fully autonomous landing with dependence on equipment or low robustness.Such as it is necessary to keep the target in the field of vision.Fully autonomous landing on moving target has been achieved based on target detection and state estimation with only on-board computation,which can also cope with temporary loss of vision.In addition,due to various interference(such as wind and load changes)in the actual working environment,and unstable or coupling features in attitude loop of quadrotor,it is fundamental to design high performance controller.The controller of attitude loop and autonomous landing control system is researched in this paper,and the main contents are as follows:The structure of the quadrotor is introduced,and the dynamics model is established,and the double loop control structure is introduced,and the position and attitude controllers are designed based on PD control algorithm.The attitude state estimation based on kalman filter is designed.The attitude is predicted by integrating the angular velocity measured by gyroscope,and is corrected by reprocessing accelerometer and vision.The effectiveness and necessity of the state estimation are verified by experiments.Two kinds of control algorithms are studied in this paper.The first is the online inertia identification based anti-disturbance control algorithm.Using a filtering process,an offline identification algorithm is extended to the online case.Using an extended state observer to estimate the disturbance,this kind of anti-interference control with self-tuning parameters has obvious advantages over PID and anti-interference control method which cannot estimate inertia parameters.Simulation and quadrotor hover experiments prove this.Model predictive control is also studied to improve the response performance of attitude system.Experiments show that the performance of the quadrotor position system is also improved.This paper uses a simple landing strategy and a mobile platform to achieve visual autonomous landing and complete the experiment.In this paper,a flow control method based on state machine transformation is studied to solve the autonomous landing problem for the complex environment and various moving targets in the practical work of quadrotor.In order to make the designed landing system adapt to the actual needs,the detection and pose estimation based on visual signs,the state estimation and trajectory planning unit of the moving platform are realized.The robot operating system(ROS)is used to manage the frequency division communication of different functional units.The landing system is simulated in Gazebo physical simulation engine to verify the effectiveness of the proposed landing system.According to the actual needs,it is very convenient to modify the landing system,such as changing the flow chart in the state machine,designing more complex state estimation and trajectory planning units,etc.The quadrotor UAV can land safely and quickly to the moving target platform in various working environments.