Autonomous Cognition And Skill Development System For Robot Service In Home Intelligent Space

Service robots play important roles in home intelligent space,and the quality as well as intelligence of the robots' service is one of the key factor which affects the home comforts and the user's feeling.As people's demand for personalized services,the number of personalized services increase with the number of devices and services in home,and existing single service conflict resolution strategy has been unable to resolve the conflicts between these personalized services.Traditional service cognition methods,including "knowledge-based" and "learning-based" methods,have achieved fruitful results in the field of robotic service cognition,but these methods are aimed at specific tasks,which results that robots lack the capacity of extended learning and self-developmental.Therefore,it is an important research content to equip robots with human-like learning and development modes,to be specific,making robots to maintain continuous learning capabilities in a complex and dynamic environment,and development themselves by simultaneously updating the knowledge base in its brain,thereby providing experience support for subsequent service cognition.This article takes the home intelligent space as the background,and designs an autonomous cognition and skill development system for robot service in home intelligent space,enabling robots to do service cognitive based on human-like learning and development models,further more detect conflicts between multi-user services and select appropriate conflict resolution strategies flexibly to meet the personal needs of every user as much as possible.The main contents are as follows.(1)Integration and monitoring for home intelligent space information.A MySQL database is designed for persistently storing data collected by the Internet of Things,and the data is divided into four major areas including users,environment,equipment,and basic information by function.Construct an ontology model to eliminate heterogeneity between data,which is regards as the upper data warehouse of intelligent space to provide data for all units below it.Develop a website for users to monitor data whenever and wherever possible.(2)Robot service cognition and skill development oriented single user.A BP(Back Propagation)neural network is construct to realize the robot service recognition.In order to avoid the tedious process of data collection and labeling,a reinforcement learning model is constructed to enable the robot to explore and learn the service cognitive by itself.Simultaneously,a development framework based on the IHDR(Incremental Hierarchical Discriminant Regression)algorithm is designed to realize accumulating experience and developing skill of robots,giving robots humanoid learning and development models and improving the scalability of its skills.(3)Further development of robot personality service cognition oriented multi-user.Provide personalized services for users based on their preferences,define the types of conflicts may arise between multi-user personalized services,and by which robot can select different conflict resolution strategies to resolve conflicts.The learning results of multi-user service cognition are updated to the IHDR knowledge base to realize the further development of the service cognition skill.Finally,the service sequence is decomposed into a sequence of atomic actions which the robot can directly execute for the robot execution.(4)Modules integration and testing of robot service cognitive system.Integrate the functional modules involved in the system,and do functional tests and performance tests in units,finally load the system into robot to test the overall system function.