Three-Dimensional Reconstruction Based On Single-Line Lidar

With the development of science and technology,the role of intelligent mobile robots in human society is becoming increasingly prominent.According to differences in working environments,intelligent mobile robots can be divided into indoor mobile robots and outdoor mobile robots.This thesis mainly focuses on the research of 3D reconstruction of real environments by indoor mobile robots.As an important tool for obtaining external information,sensors have been widely used in various intelligent mobile robots,usually divided into cameras,multi-line Li DAR,and single-line Li DAR.Cameras usually have high requirements for ambient light;Due to vertical angle measurement constraints,multi-line Li DAR cannot meet the requirements of omnidirectional mapping in indoor environments.This combines various factors such as usage environment,manufacturing cost,and construction quality to design a 3D reconstruction system based on single-line Li DAR.Compared with the same type of products on the market,this system has lower costs,a wide perspective on drawing,higher quality and efficiency,and solves practical engineering problems.It has a certain practical value.The main work of this thesis is as follows:(1)Design and build a 3D scanning system based on single-line Li DAR.A 3D scanning system design scheme based on single-line Li DAR is proposed to address the issues of planar scanning of single-line Li DAR and vertical angle measurement constraints of multi-line Li DAR.The 3D scanning system components of single line Li DAR,motor,and communication circuit based on STM32F103RCT6 are completed.(2)Preprocess the point cloud data collected by the system.Align the original point cloud data collected by the 3D scanning system with a timestamp first.To address the problem of large amounts of point cloud data and the presence of noise,noise reduction filtering is carried out.Different methods are used to filter obvious side leakage points,isolated points,and noise mixed with correct points,providing accurate data information for subsequent 3D reconstruction of real environments.(3)Realize the 3D reconstruction of the real environment by the 3D scanning system in a single perspective state.When constructing a map for a specific area,the reconstruction system is in a stationary(single view)working state.Firstly,linear interpolation algorithms are used to calculate all point cloud data in a single area,and then Euler angles in rigid body transformations are used to perform 3D reconstruction of the real environment.The experimental results show that the reconstruction effect of the system is good,with small errors,and meets the accuracy requirements for engineering applications.(4)Complete the 3D reconstruction of the real environment when the 3D scanning system is in a multi-view state.The reconstruction system is in a motion(multi-view)working state when constructing a global map.Before point cloud registration,an improved curvature method is used to extract point cloud feature points.The displacement provided by the wheel odometer,the rotation angle provided by IMU,and the data calculated by the NDT point cloud registration algorithm are weighted and averaged to achieve a 3D reconstruction of the real environment from multiple views.By comparing the experimental results of rooms and corridors,it can be concluded that the reconstruction effect of the 3D scanning system is good,and it can reflect the 3D features of the real environment with small errors.