The Appliance Of Natural Language Understanding System On Lake Water Clean-up Robot

Lake clean-up robot is a new environmental protection robot, and it provides a new way to treat artificial lake. With the Natural Language Understanding system, it becomes easy to control the robot. Based on the theory of NLU, according to the character of water clean-up robot of the lake, this thesis researches the application of natural language understanding.Firstly, we collect and analyze a great deal of commanding sentences. Then, according to the present NLU theories, we design the system. The center word of the sentence is verb, and it marches the Case Grammar. According to the character of robot itself, we bring forward a new way to advance it. By analyzing commanding sentences, we find that all the sentences can be divided into two classes, and the verbs in each class have the same character, so this thesis divides all the verbs in the system into two classes, the first class and the second class.Secondly, the system includes three parts, dictionary analysis, syntax analysis, and semantic analysis. Because of the divide of verbs, Syntax analysis becomes more easy and quick than the way of up-to-down or down-to- up, and the accuracy and efficiency of the system syntax analysis is advanced. The first step of semantic analysis is to analyze the frame of the semantic. We apply Conceptual Dependency Theory to solve the problem that the same meaning has different ways of expression. We can get the object of movement from the verb of the second class that we define in the system. If there is only one such verb, we can get two cases, movement and movement object. The verb of the first class couldn't get the object of movement, but such a sentence always includes a verb of the second class or noun to speculate the object of movement. The numerals and classifiers in the sentence include the related information of timer movement. According to the standard measurement unit of the system, it can do the related operation and get the added cases that we need. The second step is to change it into command parameters, and change the semantic frame into the parameter set that the robot orders need.Finally, using the sentences in the corpus, we test the NLU system. According to the problems of the system, we put forward the resolving methods.