As a kind of multi rotor UAV,four rotor UAV has the advantages of simple structure,vertical take-off and landing,and is widely used.Based on the application research of machine vision,this paper applies machine vision to the four rotor UAV,and carries out the research,design and debugging of UAV Based on machine vision.The whole system platform consists of flight control platform and vision unit.The flight control board adopts STM32 processor and onboard gyroscope and other sensors,which is responsible for real-time estimation of the UAV’s position and attitude information and driving the rotor operation,providing the basis for the UAV’s stable flight.Then,the flight principle of four rotors is studied,and the dynamic model of the system is established by using Newton mechanics method,and the attitude estimation algorithm based on complementary filter is designed for the established model,which can calculate the attitude information of current UAV in real time.The vision unit adopts Samsung 4412 processor,external USB camera,transplants opencv library,studies the target detection algorithm,extracts the target position by using ecological processing method,and sends it to the flight control platform through serial port for visual inertial navigation processing.Finally,the final tracking flight can be realized by controlling the motion of the aircraft through the position and attitude control algorithm.Finally,the above test algorithm is run in the whole system for debugging and analysis.The vision unit can accurately identify the target object,and the recognition speed is fast,which meets the real-time requirements of UAV pose estimation.The flight control board receives the position of the target pixel from the vision unit,and estimates the position and attitude.It can effectively and quickly calculate the attitude information of the UAV,with high accuracy,fast tracking and no obvious lag phenomenon.When the system is powered on for flight test,moving the target object,the UAV can track the flight accurately and quickly. |