Font Size: a A A

Research On Route Natural Language Processing Method For Robot Navigation By Asking The Way

Posted on:2018-12-20Degree:MasterType:Thesis
Country:ChinaCandidate:Q ChenFull Text:PDF
GTID:2348330518457610Subject:Engineering
Abstract/Summary:PDF Full Text Request
Navigation technology is one of the core technologies of mobile robot.Robot navigation by asking the way through natural language is one of the ideal methods to realize the robot navigation task.In order to use the natural language to controll robot to complete autonomous navigation,this paper proposes a semantic extraction method based on semantic role labeling(SRL),which is used to enhance the understanding of route natural language for robots.Firstly,a non-restricted route natural language corpus is constructed.Then,based on the thorough study of the natural language corpus,this paper puts forward 8 lexical chunks that contain the key navigation information.Manually tagging the original corpus by using the 8 lexical chunks,and then manually tagging the semantic role labeling corpus based on lexical analysis.Based on the corpus,the conditional random field is used to analyze the lexical chunks and the semantic role labeling based on lexical analysis.Secondly,the original corpus is manually annotated to get the dependency parsing corpus,and on the basis of this,a semantic role labeling corpus based on dependency parsing is obtained.Based on the corpus,the conditional random field is used to carry on the dependency parsing and semantic role labeling based on dependency parsing.Then,a semantic role labeling method based on lexical analysis and dependency parsing is proposed.The experimental results for the precision,the recall and the F1-measure are 98.22%,98.48% and 98.35%,respectively.Thirdly,based on the result of semantic role labeling,this paper proposes a route navigation information extraction method based on semantic role labeling.A complete route divided into several route units,according to each POS and chunk annotation results,dependency syntax tagging results and semantic role labeling,to determine the critical route information,the complete virtual navigation route is finally obtained.Finally,based on the method of route natural language processing(RNLP)proposed in this paper,the robot Nao is used to carry out the experiment of robot navigation by asking the way.
Keywords/Search Tags:robot navigation by asking the way, semantic role labeling, chunking, dependency parsing, route natural language processing
PDF Full Text Request
Related items