The Research On Mobile Robot Mapping Based On Fusion Of UWB And LiDAR

Mobile robot has played an important role in many fields such as industry,healthcare,and services with the rapid development of robotics as well as theories such as artificial intelligence.Being able to construct an environment map is an important part of almost every autonomous mobile robot,especially if they have to perform specific tasks in unknown environments.Therefore,the accuracy and quality of the map directly affects the ability of the robot to perform the task effectively.Since the map built by low-precision 2D Li DAR is incomplete,inaccurate,and not capable of constructing map in complex environments.The traditional map building algorithm that uses short-range 2D Li DAR to build large-range and low-feature maps,which leads to a large deviation of the built map from the real environment due to the incomplete information detected by Li DAR and the large cumulative error caused by the odometer running for a long time.This paper aims to build a high-precision map of a large-scale environment in a low-cost way to meet the requirements of practical industrial applications.The specific research can be divided into the following parts.Firstly,odometer information is used for locating the mobile robot and a map of the environment.The map can be constructed with the odometer information combined with distance and angle information obtained from 2D Li DAR sensors.At the same time,a platform for collecting odometer and Li DAR data is established based on the map building system.Secondly,since this paper adopts the graph optimization technique to fuse UWB sensor and Li DAR sensor information for high-precision map construction,UWB is used to eliminate the cumulative error of odometer and correct the robot's pose during the first graph optimization stage.However,UWB is affected by the NLOS error which will lead to large fusion positioning error.Therefore,this project analyzes the impact of UWB ranging error on the actual positioning and adopts two particle filter algorithm to eliminate the ranging error in order to improve the positioning accuracy and prepare for the construction of a high-precision mobile robot map.Next,a novel solution is proposed to address the accuracy problem of mobile robot map building.Odometer,UWB and Li DAR information are fused into two map optimization frameworks to achieve robust and highly accurate map building in unknown and featureless environments.The fusion of odometer and Li DAR information can build a map of the environment.However,the large cumulative error of odometry cannot be avoided when construct a wide range environment for a long time,so fusing UWB sensor information in the first map optimization can get a more accurate robot position,and thus a more accurate map can be obtained.Experimental results show that this method greatly improves the positioning accuracy of mobile robots in non-line-of-sight environments.Then,Li DAR loop closure detection can be used to determine whether the robot returns to the same place it has been visited,so the robot position information obtained from the first map optimization technique is combined with Li DAR loop closure detection information can greatly improve the mapping accuracy of the mobile robot during the second map optimization.The experimental results show that the method effectively improves the mapping accuracy.Finally,in order to improve the accuracy of loop closure detection of short-range 2D Li DAR,a method of loop closure detection based on sub-map matching is proposed in this paper.The method stitches multiple frames of Li DAR point cloud data into sub-maps for loop closure detection to compensate for the limitations of using single-frame Li DAR data for loop closure detection in low-feature environments,and the experimental results show that method greatly improves the accuracy of mobile robot map building.