| Simultaneous localization and mapping(SLAM)is the key problem that must be solved in the process of locating in unknown environment.The establishment of two-dimensional map and localization using Lidar is an important part of SLAM research.The main problem of SLAM is involved for additional map and location.After Lidar scanning unknown environment,transform relationship,including rotation and translation transformation,between the current time observation and original created map,is obtained.This paper studies on how to build and update maps and location in real time through scanning data.The main work includes:Lidar data acquisition.In unknown environment,data collection and processing are the prerequisite for further work.Firstly,the data acquisition and transmission process of lidar is studied.Secondly,the establishment of grid map is mainly studied.After analyzing the original method of building grid map,a simplified method for creating grid map is proposed.First,the two-dimensional planes are divided into equal area grids.Next,calculate the location of the grid where each scan point falls,and then save the grid in the same time.Finally,all the saved grid is shown as grid maps.Simulation results show the effectiveness of the proposed method.Branch and bound based search matching algorithm.Matching is a main method of obtaining the transformation relation between the scanning points in the current time and the map created by the original.This paper proposes a branch and bound search matching strategy without seeking the corresponding feature points.And then,the feasibility of matching strategy is proved,also the time complexity of the existing algorithm and proposed one is analyzed and compared.Also,in the search process,a search strategy based on relaxation factor is proposed to change the search step length to reduce the influence of step size on the search results.Multi frame matching algorithm based on variable point set.The two-and-two matching strategy is proposed as multi point sets matching.In order to reduce the cumulative error of two-and-two matching in multi point sets matching,an inverted matching strategy is proposed.Meanwhile,loop closure is used on scanning points.With the increasing number of frames,the accumulative errors will make the map accuracy reduced,so line extraction in the map is needed to improve the accuracy.The traditional split and merge algorithm is analyzed and improved.Implementation of real time simulation system.Visual Studio platform is used for real-time simulation.First,configure the drive files needed for the PCL library and the Lidar;Secondly,introduce and design the overall implementation plan and the flow of the real-time simulation process.Finally,the system test is carried out to realize SLAM in the unknown environment. |
