Research On Location Technology Of Indoor Mobile Robot Based On Lidar

With the development of robot technology and its popularization,indoor mobile robots have become a kind of common products,among which indoor sweeping robots are the most representative.As soon as the sweeping robot was launched,it became a popular project in the market.The indoor positioning system is the most important part of the continuously upgraded sweeping robot.Based on this background,this paper studies the positioning system of indoor sweeping robot,studies the construction of positioning algorithm based on particle filter algorithm,and optimizes particle recommendation distribution function by fusing radar matching algorithm and RBPF particle filter algorithm.In addition,a fast registration algorithm based on the linearized expression of lidar data is proposed to improve the real-time performance of the localization algorithm.The thesis is mainly engaged in the following work:Firstly,according to the structural characteristics of indoor sweeping robot,its motion model is constructed,and the sensor model of the robot is constructed,including odometer model and environmental expression model based on lidar.This paper studies the existing indoor map expression methods,and explores the construction method of raster map.Secondly,scanned data are obtained from lidar sensors.Aiming at the problem of time-consuming calculation and poor real-time performance of traditional ICP algorithm,a fast inter-frame point cloud registration method is proposed by linearizing the point cloud data.Linear expression of point cloud includes point cloud partition and line fitting.According to the structural characteristics of indoor environment,the principal direction angle of the environment is extracted from the linear expression information,which is used as the condition of fast point cloud registration.The effectiveness of the improved registration method is verified by MTLAB simulation.The experimental results show that the average time of the improved algorithm is 35% of that of the traditional ICP algorithm.Finally,the SLAM localization algorithm based on particle filter is studied,its implementation principle is analyzed,and the implementation process is deduced.The improved point cloud registration algorithm is added to the RBPF-SLAM algorithm,and the program is written and tested on the experimental platform based on raspberry pie and STM32 to verify the positioning effect of the improved algorithm.Through field composition experiments,the results show that the improved algorithm is more accurate in positioning,smoothing the boundary of the map,and better in effect.