Simultaneous Localizaton And Mapping Of Unknown Indoor Environment Based On Lidar

With the development of technology and the promotion of artificial intelligence industry all over the world,the robot industry and its basic theoretical research are paid much attention by researchers.The simultaneous localization and mapping(SLAM)technology is a core technology for the self-exploration of artificial intelligence mobile robots in many geographical environments,which has become a research hotspot in the field of robotics and has important research value and economic value.In this paper,the simultaneous localization and mapping of unknown indoor environment based on 2D LIDAR sensor is studied.This paper describes the SLAM problem based on 2D LIDAR with mathematics.Then a lightweight SLAM framework was proposed according to the characteristics of indoor scene and characteristics of 2D LIDAR data.Finally,based on the proposed SLAM framework,a SLAM system is designed.In the registration of LIDAR data,this paper firstly analyzes the factors affecting the registration algorithm of LIDAR data,and put forward the problem of large motion which cannot be solved by the existing registration algorithm.Then give a mathematical description of the 2D LIDAR registration problem.Finally,in order to solve the problem of large motion,this paper proposes an adaptive threshold line feature extraction method and a data registration method based on line feature combining density clustering and geometric principle.In the aspect of mobile robot mapping,this paper first uses Kalman filter to correct the pose of mobile robot.Then,the coordinate transformation of mobile robot in the process of the map generation is analyzed.Finally,the principle of generating occupancy grid map based on 2D LIDAR data is analyzed and deduced in detail.To verify our work,this paper carried out simulation experiments.We first built a simulation experiment platform based on MATLAB.Then a comparison experiment of the data registration algorithm is carried out to prove that the data registration algorithm based on the line feature combining density clustering and geometric principle can effectively solve the large motion problem and has better robustness than the classical algorithm.Finally,a comprehensive experiment of SLAM based on 2D LIDAR is carried out,which proves the necessity and effectiveness of Kalman filter for multi-sensor fusion,and proves the feasibility of the proposed SLAM framework.We provide the code we wrote in this research.