Mobile Landing Of Drones

With the rapid development of science and technology,domestic and foreign robotics technology is booming,and the development of drones is also changing rapidly.The drone itself contains a large amount of theoretical knowledge and engineering framework.Many domestic and foreign researchers continue to study the various aspects of drones.Regardless of its top-level drive control or upper-level application,every theoretical breakthrough or engineering improvement can provide a solid foundation for the qualitative leap of drones.In order to better expand the application of unmanned aerial vehicles,it is often necessary to extend the endurance of drones,and the autonomous landing technology of unmanned aerial vehicles plays an extremely important role.This paper focuses on researching key technologies such as relative position,relative speed,target tracking,landing strategy,etc.in mobile landing.Independently develops a whole set of research platforms,and integrates a set of mobile landing solutions and existing technologies.The slight improvement is mainly divided into:(1)Self-developed flight control,specific design of software and hardware,useage of gradient descent for attitude calculation,cascade ADRC attitude control and position control,which increase the anti-jamming performance of the drones,and finally realize the fixed-point and cruise-by-waypoint functions.The development of the software and hardware of the landing platform controller,real-time transmission of the landing platform’s attitude,position and speed and other information related data to the drones,to provide a certain basis for the study of mobile landing.(2)Apriltags visually recognizes two-dimensional code tags,provides the relative position information of the drone and the landing platform,as well as its related data processing,and assists the landing of the drone.(3)The overall strategy of mobile landing is briefly summarized as follows: safety detection before take-off,and then tracking after flying to a preset safe altitude.When the vision processor recognizes the QR code,it enters the ready-to-land mode and detects the status of the landing platform in real time during landing to ensure a safe landing.(4)In view of the limited recognition frame rate of the vision processor and the speed of the tracking target changes too much,which leads to the failure of recognition.To solve this problem,the position estimation algorithm is improved.Besides,a position velocity estimation method combining vision and GPS data is proposed.The velocity difference and GPS position accuracy factor are used to carry out variable weight complementary filtering,and the better relative position and relative velocity are calculated.Aiming at the problems of inaccurate acceleration differences under high-speed conditions and slow tracking with small errors in the commonly used tracking methods,a tracking controller with self-anti-interference structure is proposed to obtain a more accurate tracking speed through ADRC.Therefore,acceleration difference information makes up for the problem of large acceleration zero offset data inaccurate at high speed,and the non-linear feedback method with small error and large gain increases fast tracking of moving targets with large acceleration,increasing the robustness during tracking.