Research On Robot Cognitive Strategy Oriented To Intelligent Service

The application of service robots has made significant progress in recent years,but its intelligent service capabilities are still at low level.At present,service robots are concentrated in simple application fields such as cleaning and education assistance,lacking ability to perform advanced services.The main constraints includes:lacking integration of heterogeneous environment knowledge of environmental recognition,which makes various types of information expressed in isolation and cannot reflect the overall characteristics of environment;Also,the execution of services requires the acquisition of in-depth service knowledge including the functional attributes,operation methods and task relationship chains of the items,but there is still a lack of means to obtain such knowledge;Simultaneously,robots perform corresponding services in accordance with pre-set task rules,lacking ability to adjust services to the environment.In response to the above-mentioned problems of the robot service process,this paper proposes a robot cognitive strategy with service as the core.Its main purpose is to enhance the intelligent service capabilities of robots,break through the limitation that service robots can only perform simple tasks and reduce manual participation,also avoid users to service the robots before enjoying robot services,so that robots have autonomous service capabilities.The main research contents are as follows(1)Aiming at the problem of insufficient semantic relevance in the knowledge cognition of the robot environment structure,an environmental semantic knowledge system of "indoor functional area-scene--item position association--item instance recognition" is proposed.Firstly,the indoor functional area is identified according to the ResNet model,and the attribution relationship between the room and the object is constructed.Then,in order to reflect the characteristics of the position of the items in the functional area,the items are divided into marked items and service items,and the visual relationship detection algorithm is used to express the position relationship between items,which fully expresses the information of the item location structure in the environment.Finally,in order to reflect the distribution of items more accurately,the instance retrieval algorithm is used to distinguish different items,complete the cognition at the instance level,and map it to the language space together with other information to form the unity of visual attributes in the environment(2)Aiming at the problem that the in-depth service knowledge required by robots depends on manual construction,an Internet-based service information expansion method is proposed.First,a topic analysis module that combines VSM model,PageRank algorithm and topic co-occurrence words is proposed to improve topic-focused crawlers,improving the accuracy of web text acquisition by crawlers.Since there are a large number of similar texts in the family task knowledge of different knowledge sources,the documents with similar service task information are clustered together through the text clustering algorithm to extract representative knowledge and filter out repeated information.Then the service tasks obtained is saved in a form suitable for robots to understand.Also,in order to extract the item attribute information in the natural text,a sequence labeling model is used for training,and the item attributes are extracted autonomously based on this model.(3)Aiming at the problem that it is difficult for robots to adjust services autonomously,a knowledge growth algorithm based on scene feedback information is proposed.Knowledge growth is divided into the semantic growth of a single task and the growth of association rules among multiple tasks.Semantic growth occurs when the task item is missing,and other items with similar attributes are used to replace the missing items to form a new task execution step.Rule growth is to dig out the execution relationship between tasks through the association rule mining algorithm,thereby extracting the user’s daily habits,so that the robot has the ability of personalized service.The cognitive strategy proposed in this paper is closely integrated with robot services,laying a foundation for robots to further improve their intelligent service capabilities.By designing robot service cognition experiments to verify the method in this paper,it proves the feasibility of the method in this paper,and it has important theoretical value to promote the practical process of service robots.