Font Size: a A A

Localization And Mapping Of Indoor Mobile Robot Based On Multi-source Data Fusion

Posted on:2018-06-12Degree:MasterType:Thesis
Country:ChinaCandidate:D Y ZhangFull Text:PDF
GTID:2428330566951557Subject:Control theory and control engineering
Abstract/Summary:PDF Full Text Request
The positioning and environment perception of mobile robot in indoor environment is the basis for robot to realize autonomous movement.Based on the laser sensor positioning and map construction system with its high precision measurement,anti-interference,good robustness and other advantages quickly become a hot research.In this thesis,a twodimensional laser scanner based on the collection of point cloud information,integrated odometer and inertial navigation pose data development of a set of high-precision indoor real-time map construction system to achieve indoor three-dimensional spatial information efficient collection.The main contents of this thesis are as follows:In order to realize the high-precision fusion of two-dimensional point cloud data and pose data,that is multi-source data fusion,the clock synchronization based on precision time protocol(PTP)is developed and the synchronization time precision is 600 ns in the local area network.Through the odometer and inertial sensor combination navigation scheme,it provides the mobile robot pose information,and use the integrated navigation method to calibrate the odometer.Based on the time synchronization protocol,the point cloud data and pose information are merged,the scanning matching algorithm is applied,and the indoor 3D point cloud map is constructed in real time,and the map is displayed in real time.The advantage of the map building system is to improve the efficiency of indoor point cloud data collection.Due to the cumulative error in the pose data and the trajectory estimation algorithm error,when the system is run in a large scale environment,the construction map has a certain deviation from the actual terrain.In order to eliminate the above deviations,this thesis proposes a differential odometer and target recognition algorithm to correct the 3D point cloud map.This thesis develops the data acquisition subsystem with STM32F407 microprocessor as the core,and realizes the preparation of the upper interface of Qt platform.The experimental results show that the map construction system can efficiently collect the indoor spatial information and construct the globally consistent 3D point cloud map offline.
Keywords/Search Tags:Data fusion, Clock synchronization, 3D point cloud map, Scan matching, Target recognition
PDF Full Text Request
Related items