Research On Chinese Speech Instruction Parsing Approach For Home Service Robot

As the two important research fields of artificial intelligence,Intelligent Robots and Natural Language Processing have become the most active fields in artificial intelligence research.It has become an important trend in the study of robots that interact with the server robot through natural language.Based on human service robot and the artificial intelligence technology,combined the domestic and foreign research status in this field,this study analyzes the natural language instruction by using natural language understanding,knowledge base,combination category grammar and man-machine dialogue technology.For the robot can be carried out within the control command to start the study.Firstly,it introduces the theory and related technologies involved in the home service robot Chinese instruction parsing,including the basic theory of natural language understanding,the core technology of knowledge base,the combinatory categorial grammar and man-machine dialogue technology etc..Through the description and analysis of these knowledge for the follow-up study to establish the theoretical foundation.Secondly,in the part of knowledge base,it is proposed to use the predicate logic to transform the unstructured natural language into structured knowledge and construct the robot knowledge base.In the part of segmentation,In order to improve the quality of word segmentation,three different combinations are designed based on the conditional random field model,and the Chinese word segmentation is carried out..At the same time,it describes the lexical structure and combination rules of combinatorial category grammar,and extends the dictionary and combination rules according to the characteristics of home service robots.Thirdly,the design of a Chinese instruction parse method.The key of the parsing method is the probability of combinatory categorial grammar model generated by parameter estimation and dictionary of probability models for combinatory categorial grammar learning,and the algorithm of dictionary generation is analyzed and improved for the performance of the parser.In addition,the process of dealing with missing and ambiguous information in human-computer dialogue is given.On the other hand to ensure the Chinese command can be smoothly parsed and converted into control commands inside the robot can perform.Finally,the experiments are used to evaluate the Chinese word segmentation technique and the Chinese speech instruction parsing method.The results of experments are analyzed by validity and feasibility verification.