Distributed SLAM Research Based On LiDAR

Simultaneous Localization and Mapping(SLAM)is one of the major research issues in robot navigation,which requires sensors to collect data.It is evitable that data from sensors also creates errors out of algorithm control.The errors will accumulate due to the increasing working condition of SLAM.SLAM algorithm revises location information frequently to minimize the errors,which in turn makes it less efficient after processing the calculation.This thesis discusses about how the current SLAM algorithm appropriately uses target func-tions or collect data based on probabilities to solve the problem after comparing several different SLAM algorithms.They all increase the efficiency by decreasing the calculation load.Single com-puter is less efficient than multiple computer when it comes to heavy calculation load by SLAM.This theis brought up distributed SLAM algorithm for multiple computer calculation which in-cludes:· The distributed SLAM algorithm is obtained by analyzing and comparing various SLAM algorithms and selecting a suitable lidar-based SLAM algorithm to extend.· The feasibility of the distributed SLAM algorithm is theoretically proved.The distributed SLAM algorithm will block the data,and the amount of data that a single machine needs to process will be reduced.Then the efficiency of the corresponding algorithm will be im-proved.In the SLAM problem,the pose solving problem can be calculated by the algorithm under the distributed computing framework.· The feasibility of distributed SLAM is proved by experiments.In this thesis,the existing public data set is used for verification.The distributed SLAM algorithm of two or three nodes is used to conduct experiments,and the running time of the corresponding algorithm is counted and the accuracy of the result is calculated.It can be seen from the experimental results that the distributed SLAM algorithm using two nodes and three nodes for calculation will have different degrees of improvement in operational efficiency.Therefore,this thesis can get a conclusion that the distributed SLAM algorithm can be used to solve the problem of low efficiency of the SLAM algorithm due to the large amount of computation.Moreover,the current single-node SLAM algorithm solves the problem of low operational efficiency to a certain extent in a corresponding manner,but these algorithms have high hardware requirements for the device when implemented.When the distributed SLAM algorithm is implemented,the hardware requirements will be greatly reduced.The implementation cost of such an algorithm will be reduced.Furthermore,the distributed SLAM algorithm can process data from multiple LiDARs,and the application of multiple LiDARs can help reduce the loss of precision caused by the presence of scanning dead zones in LiDAR.