Research On Vision-based Moving Target Tracking System For Four-rotor Aircraft

The mechanical structure of the four-rotor aircraft is simple, with the characteristics of vertical landing, autonomous hover and flexible control, which can be applied to various flight posture and flight profiles, and plays an important role in military and civil fields such as riot control and traffic monitoring. Therefore, the research on the vision-based moving target tracking technology of four-rotor aircraft has important research significance and broad application foreground.This paper firstly introduces the flight principle of four-rotor aircraft, analyzes the coordinate transformation between the ground coordinate system and the body coordinate system,and establishes the mathematic model of four-rotor aircraft. Because of the difficulty of detecting and tracking moving objects under dynamic scenes, in order to improve the robustness of the algorithm and reduce the complexity of the algorithm, this paper adopts a color-based target recognition algorithm, and combines the Camshift with the Kalman filtering algorithm as the tracking algorithm, the accuracy of the algorithm is verified by experiments In the case of color disturbance and target occlusion. In order to improve the practicability of the this system, a control panel with STM32F405RGT6 as the core is designed, connect the control board to the four-rotor aircraft and the vision-based moving target tracking system for four-rotor aircraft is constructed. A PD controller with dead zone is designed, real-time acquiring the position deviation of the target center and image center in the image plane, as input of the controller, the control data is computed, and the closed-loop controller design based on image sequences is accomplished. Finally, the design of the system program and controller is completed, and it is successfully transplanted into the embedded development platform.In order to verify the performance of the parts of the system, the AR.Drone four-rotor aircraft is used as the flight piatform, connect the control board to the AR.Drone four-rotor aircraft for tracking flight experiments. The experimental results show that the system has good target recognition and tracking ability, and verifies the effectiveness of the system.