Research On Flight Control System Of UAV Based On Computer Vision For Intelligent Steward

In recent years,with the emergence of a series of landmark events such as Alpha Go and driverless cars,artificial intelligence has gained people's high attention.The combination of artificial intelligence and other traditional fields will bring new development opportunities to these disciplines.Artificial intelligence butler is a dream of human beings,representing a peak of artificial intelligence,but humanoid robot butlers who can enter ordinary families still need a long research and development process.We intend to use a combination of artificial intelligence,computer vision and drone technology to develop a lightweight intelligent butler system based on small drones.This paper focuses on the development of the flight control system of the drone.The research and development work completed is as follows:.The DDC technology was used to develop the ground command system of the drone.The circuit interface between the computer and the ordinary UAV remote control was realized by using the USB interface circuit module and the 4-channel D/A conversion + 8-channel GPIO circuit module.The remote control was not changed.Under the premise of firmware and onboard control circuit hardware,direct digital control of the drone on the ground desktop computer is realized.The indoor positioning technology of UAV based on computer vision is studied.The UAV positioning technology based on PNP principle and the UAV positioning technology based on matching feature points are discussed.It is found that although the two have higher positioning accuracy,The former is marked in the venue in advance,and the latter is inferior in real-time.For this reason,the paper proposes a fast feedback positioning method for drone based on 4 quadrant matching feature points.The method can quickly determine the relative position of the drone by comparing the change of the relative position of the matching feature points in the current image and the reference image,and realize the fast and accurate positioning of the drone.Further on this basis,the positioning and cruising flight of the drone at the designated point in the room are realized.The computer-based UAV indoor safe flight and obstacle avoidance technology is studied.The detection and obstacle avoidance of the active personnel are realized through face and pedestrian detection.At the same time,an inherent object detection and obstacle avoidance method based on 3D reconstruction is proposed.The method acquires a sparse point cloud of a scene by means of SFM,and constructs a three-dimensional feature point library by means of these sparse point clouds.The feature points in the current image are further matched with the feature points in the 3D feature point library.By matching the successful 3D feature points,the UAV world coordinates are solved,and the distance from the object point is calculated to realize automatic obstacle avoidance.Developed a set of intelligent butler flight control software system based on drone,which can easily carry out experimental debugging and parameter setting for each module and function of the system,including remote control hardware control,image acquisition,image processing,drone control,fixed point With cruise flight control,active personnel and intrinsic object detection and avoidance.Experiments show that the software and hardware intelligent control system developed in this paper basically meets the needs of intelligent butler systems based on drones.