Service robot instruction analysis method based on key verb feature full-density transmission

Abstract

The invention discloses a service robot instruction analysis method based on key verb feature full-density transmission. The method comprises the following steps: working in a known application environment, constructing a database of various service instruction texts under various candidate services of a service robot, and preprocessing the service instruction texts; dividing the database into a training set and a verification set; performing word segmentation on each service instruction text by adopting a Steak text analysis tool to obtain context features of the key verbs; building a full-density bidirectional long-short-term memory network model, training model parameters by using the training set, and performing hyper-parameter debugging by using the verification set; and processing and extracting a to-be-tested text instruction to obtain the type of the candidate service and instruction semantics under the candidate service. According to the method, a service instruction analysistask is converted into two sub-tasks of frame classification and frame filling, and core elements in a service instruction text with redundant information are accurately captured and extracted and used for assisting a robot in understanding complex instruction intentions.

Description

technical field [0001] The invention relates to a robot instruction analysis method belonging to the field of robot control, in particular to a service robot instruction analysis method based on full-density transfer of key verb features. Background technique [0002] In the process of interaction between humans and intelligent service robots, the use of natural language expressions is beneficial for users to issue task instructions. Since there are often redundant information or insufficient expressions in user task descriptions, robots need to understand tasks, that is, to parse open-ended complex instructions into a structured internal representation for subsequent execution. Frame Semantic Representation is a reliable machine internal representation method, which transforms the task of instruction parsing into two subtasks of frame classification and frame filling by constructing a frame structure covering the task content. [0003] At present, the two main methods for ...

AI Technical Summary

Problems solved by technology

The neural network method only relies on word vector features for semantic analysis, ignoring the dependence of the overall task intention on different word vectors, making it difficult for the neural network to focus on key verbs containing core task information

Images