On-line Construction And Updating Of Knowledge Base For Home Service Robots

Service robots have good application prospects in the fields of family services,assistance for the elderly,disaster relief,and special services.To make robots more intelligent,the construction and update of the home service robot knowledge base has become very important.This paper identifies and locates objects in the home environment through binocular vision,preparing for the construction of the knowledge base.This paper introduces the family service robot's perception of the home environment through vision,and proposes a method based on Yolo neural network for fast object recognition and localization.Then,the service robot knowledge base is constructed for the home environment ontology,and the ontology construction method of the service robot is proposed,which is divided into the environment model part and the service model part.An ontology construction tool based on home environment and service is designed.The small part of the model construction requires a small amount of manual participation to record the spatial distribution of objects in the home environment.In the service part,a model with decision-making function is proposed,which has a learning function,which saves a lot of repetitive work during construction,And adding a method of automatically building a task automaton,and uses the Baxter robot to step through the task sequence acquired by the crawler program.Utilize binocular vision to detect and record status,and use state space method to save time when completing tasks.Through the self-set experiment of moving objects in the self-set nine-square grid,it is verified that this method not only effectively reduces the update operation,but also saves time and space.By experimenting in a complex environment,using the state space method combined with the task automata to perform tasks,the robot is prevented from performing the first step each time.