A Vision Based UAV Pesticide Mission System

According to the requirements of the 5th International UAV Innovation grand prix (UAVGP 5th), a vision based UAV target recognition and task execution system is designed in this paper. By analyzing the competition task, we decompose the whole system, which including the flight control system, the structural system, the navigation system, the data processing system, the power supply system and the visual image system. The link relationship of the system are complex, so we analysis the system from the power supply system diagram, data interface, system signal flow angle etc. In order to get the accurate navigation and positioning data, we design a multi sensor fusion algorithm based on EKF. Aiming at the problem of flight control and image recognition, flight control design based on PID algorithm and a robust image recognition algorithm are developed. For the entire task execution flow, we design rules based on task execution flow. The experimental results show that the system has accurate positioning, target recognition and complete task flow process.