Service Task Generation And Cognitive Model Building Based On Intention Recognition In Home Environment

Service robots are more and more used in people's family life.In order to meet the increasing needs of users,robots not only need to be able to accept simple service instructions,but also need to improve the quality of people's lives by increasing the level of intelligent services.In order to enable robots to provide users with active and intimate services,this paper proposes service task generation and cognitive models based on intent recognition on the basis of semantic ontology.After acquiring the data resources of the service environment,expand the service cognitive knowledge base of the robot by combining the experience knowledge and reasoning rules in the home field and the service resources on the network,and integrate the sensor information to generate a complete text for the user's service intent recognition,and make a deep understanding of service tasks under the premise of confirming that users need services,so as to improve the robot's service cognition level.The main research content of this article is divided into the following four parts:Firstly,in order to be able to synthesize a variety of environmental information for service analysis,a multi-information fusion ontology knowledge database and service rule database are constructed.The ontology knowledge base constructed using the ontology description language OWL comprehensively covers the items and property attributes,environment and user information involved in family life as much as possible.At the same time,the service rule base constructed using the semantic Web rule language SWRL integrates various types of environment,user,and item information,and uses its logical reasoning capabilities to perform inference generation of service tasks.Finally,in order to facilitate the use and update of the pre-built information database,it is parsed into the form of a dictionary according to different structure trees to provide a unified interface for subsequent service cognition.Secondly,in order to make up for the problem of insufficient information in the self-built knowledge base,web spider technology is used to crawl the cognition of tasks related to home services on the network This article constructs the overall framework of a web spider based on the Requests module,obtains the details of service topics and task executions,stores them persistently in a relational database,and updates information through dynamic management mechanisms.At the same time,the rules of the service rule base are simplified using the rule inference task generation algorithm,and stored as a dynamic rule decision table through the database,which improves the inference speed and simplifies the final inference results.Thirdly,a service intent recognition module based on text generation is set up to solve the problem that the supplementary reasoning rules involve incomplete service types.By extracting the underlying sensor data into keyword expressions,adding keyword extraction and attention mechanisms to the semantic control long-term and short-term memory network,and using SC-LSTM networks based on improved semantic control to recombine keywords into complete utterances.Then TextCNN is used to determine whether the user needs service under this condition.Finally,categorize service tasks.Deep semantic extensions are provided for services that involve only a single item to facilitate service task understanding.For service tasks that involve multiple items or need to perform multiple steps,the noun vocabulary involved in the task is identified by means of part-of-speech tagging,and then the attribute of the items in the semantic ontology is extended to realize the precise search and positioning of the items by the robot.The method proposed in this paper is loaded into the ROS experimental platform that has been constructed in the laboratory to verify the realization of service cognition for different tasks.