Multi-rotor UAV Trajectory Tracking Based On Dynamics And Monocular Vision

Due to its structural characteristics,multi-rotor UAV has been paid more and more attention and applied in military,civil,agricultural and industrial fields.For these applications,the related research topics of various multi-rotor UAVs are also bred,such as the rapid grasp of the manipulator of multi-rotor UAVs,the suspension transportation of multi-rotor UAVs,and so on.The research of multi-rotor UAV trajectory tracking is the basis of these topics.With the changing application environment and the increasingly complex tasks,people put forward higher requirements for the autonomy and tracking accuracy of multi-rotor UAV trajectory tracking.At present,the trajectory tracking of multi-rotor UAV mainly uses non-dynamic control method,and uses GPS to provide positioning information.However,considering the low control accuracy and poor maneuverability of the non-dynamic control method,and when indoor GPS signals are invalid or ineffective,the GPS positioning method can not make the multi-rotor UAV better complete the trajectory tracking task.Based on the above problems,this paper designs a trajectory tracking method based on multi-rotor UAV dynamics and monocular vision.This method can ensure that the multi-rotor UAV can complete the trajectory tracking task independently in the room without GPS signal or weak GPS signal.Firstly,the complexity of dynamics of multi-rotor UAV is analyzed,and the inappropriateness of the over-complex model in controller design is explained.Based on this,a reasonable dynamic model with complexity is established.At the same time,the generation mechanism of internal forces and moments of multi-rotor UAV is analyzed by taking X-type Four-rotor as an example.The corresponding force and moment distribution model is given,which provides a more complete model for the design of trajectory tracking controller.Then,the trajectory tracking problem of multi-rotor UAV is decomposed into two modes: attitude tracking and position tracking.The theoretical design of trajectory tracking controller is completed by using the method of PD plus feedforward.Then,from the point of view of Engineering implementation,the structure of the whole trajectory tracking controller is simplified to reduce the amount of code in the process of controller implementation,and then the force and moment control of multi-rotor UAV is converted to the speed control of each rotor motor by using the allocation model.At the same time,in order to obtain the relevant dynamic parameters of the multi-rotor UAV,the relevant experimental and measuring equipment is built through investigation,and the calibration method of the relevant parameters is explained.In addition,in order to facilitate the engineering realization,the motor speed mapped by the motor thrust is converted to the pulse width by using the electromechanical dynamic parameter measuring equipment built in this paper,which realizes the real engineering control of the multi-rotor UAV from the theoretical control to the actual control.Considering the complexity of designing and building flight control hardware by ourselves,this paper realizes the engineering deployment of the bottom controller of trajectory tracking with the help of open source flight control framework PX4 in open source hardware Pixhawk.Based on the bottom controller of trajectory tracking,this paper completes the design of pose estimation algorithm for multi-rotor UAV indoor by using the integrated IMU of monocular camera and flight control,and the framework of monocular vision mileage calculation method with tight coupling and non-linear optimization.In order to further verify the feasibility of trajectory tracking algorithm and visual positioning algorithm,the X-type Four-rotor was built,and the related dynamic parameters were measured by the dynamic parameters measuring equipment built in front of it.Then,in the motion capture system,the feasibility of trajectory tracking bottom controller and visual inertia mileage calculation method is verified experimentally.It is shown that the bottom trajectory tracking controller designed in this paper has strong trajectory tracking ability and tracking accuracy,and the monocular vision inertial mileage calculation method can provide better position and attitude estimation for multi-rotor UAV.Finally,the trajectory tracking experiment proves that the whole trajectory tracking algorithm based on dynamics and monocular vision can ensure that the multi-rotor UAV can complete the autonomous trajectory tracking task better.At the same time,the trajectory tracking method proposed in this paper can greatly promote the research of multi-rotor UAV’s rapid manipulator grasping and suspended load transportation.