Research And Implementation Of Key Technologies For UAV Obstacle Avoidance Flight System

With the rapid development of multi rotor UAV technology and the gradual decline of cost,UAV is widely used in various industries.With the expansion of application scenar-ios,the flight environment of UAV is becoming more and more complex.The traditional autonomous flight scheme of UAV Based on GPS positioning system is not suitable for the environment with complex obstacles,which requires manual operator to control the UAV Therefore,the research content of this thesis is how to avoid obstacles autonomously and complete the scheduled flight mission in complex environment when UAV uses lidar to sense environmental information.This thesis focuses on the global path planning and local obstacle avoidance of UAV autonomous obstacle avoidance system based on lidar.This thesis mainly consists of the following aspects:(1)Firstly,according to the functional requirements of UAV autonomous obstacle avoidance flight,the software and hardware structure of obstacle avoidance flight system is designed.In this thesis,the whole system is divided into two parts:UAV end and ground station end.At the same time,the software is divided into map building module,global path planning module and local obstacle avoidance flight control module.These software modules are combined with hardware devices to complete obstacle avoidance flight function.At the same time,the hardware models of UAV,lidar and so on are given.(2)Aiming at the problem that obstacle avoidance flight test of UAV is inconvenient and has potential safety hazard,this thesis builds a set of UAV Simulation Test Platform Based on gazebo.This thesis mainly solves the problem that the performance and control interface of UAV and lidar hardware used in the gazebo platform is inconsistent with the hardware in the real environment.At the same time,it also builds the flight test scene in the simulation environment,which provides a convenient simulation test environment for the follow-up research of this thesis.(3)For the global path planning problem,this thesis implements the global path plan-ning module based on RRT algorithm,and analyzes the shortcomings of RRT algorithm in this scenario.An improved method of dynamic expansion parameters based on the last expansion result is proposed.By dynamically modifying the expansion parameters,RRT algorithm can adapt to the different environmental characteristics of open space and ob-stacles,and select the relatively optimal expansion strategy,so as to reduce the planning time.In addition,an improved method based on search tree reuse is proposed.When mul-tiple end-to-end path points need to be planned,the time of repeatedly exploring the same area can be saved by reusing the last planned search tree,so as to speed up the planning speed.Finally,a path optimization method is proposed,which can alleviate the problems of too many inflection points and detour.The effectiveness of the improvement is verified by comparative experiments.(4)Aiming at the problem of local obstacle avoidance,this thesis implements the local obstacle avoidance flight module based on 3dvfh algorithm.Based on the fact that the UAV can sense the obstacle information 360° through the lidar in this system,so the UAV does not need to turn the nose to the flight direction in the flight process,the cost function of 3dvfh algorithm is changed.In addition,because the original algorithm does not give the specific scheme to determine the high and low threshold in the binarization of two-dimensional histogram,this thesis proposes a threshold determination scheme based on statistical method.Finally,the local obstacle avoidance flight module is tested in the simulation environment and the actual environment,which proves the correctness and effectiveness of the module.