Research On Flexible Reasoning And Task Planning Of Service Robot Based On Intelligent Space

With the rapid development of service robot technology,people's demand for robots is no longer satisfied with the functions of entertainment and escort.How to make robots actively provide humans with more reasonable and intelligent daily services in a dynamic and complex home environment has become a topic in the research field of robots.Although intelligent space technology started a new way for the field of robot environmental perception and service cognition,home service robots often ignore the user's own habit changes and dynamic body changes when inferring service tasks.Then,the multi-attribute historical data of service tasks are often not taken into account when robots plan multi tasks resulting in unsatisfactory service results.In order to further improve the human-computer interaction ability of the robot in the environment of human-computer integration and enhance the satisfaction of users with the service results,it is necessary to build a reliable user feedback system of home service robot.Based on this,a robot service task reasoning system with reasonable user context information as the core and a reasonable multi-tasking planning strategy are designed,so that the robot can actively provide users with services that meet actual needs.The research work of this thesis is mainly composed of the following parts:(1)A deterministic service task reasoning method of home service robot is designed based on the combination of multi classification algorithm and user dynamic feedback system.Firstly,based on the historical service data of users in the home intelligent space,a service task rule base is established and stored in the MySQL database for sharing and operation.Then,the advantages and disadvantages of machine learning multi classification algorithm in service reasoning are analyzed.Aiming at these disadvantages,an improved k nearest neighbor algorithm is proposed to dynamically serve users according to their context information in the home environment.Finally,the user feedback mobile app and the web service system of the computer are developed as the home intelligent space human-computer interaction system.Users can dynamically add and modify the service rules.(2)This thesis studies the uncertainty reasoning method of user service tasks based on Multi-Entity Bayesian network.User intent recognition is affected by many uncertain factors in the dynamic environment of the home,including user factors and environmental factors.Aiming at the above problems,the probabilistic reasoning of Multi-Entity Bayesian Network is used for intent recognition.The first-order predicate logic expression ability and multi-entity fragment of Multi-Entity Bayesian Network are used to describe the uncertain information of dynamic situation and construct the complex family situation model.Then,the uncertainty service reasoning of users is carried out in the complex and dynamic environment by combining the Bayesian reasoning algorithm.(3)Based on the reasoning results of the service reasoning system,the multi-task planning problem of service robots is studied.Firstly,model and analyze the attribute diversity of home service tasks.Then,based on the historical data of multiple attributes of each service task,a decision theory planning method is introduced to perform initial task planning for multiple tasks that the robot needs to perform.Finally,the real-time planning of robot tasks based on the principle of dynamic task decomposition is designed,and the problem of cooperative execution of robot multi tasks is studied.(4)Aiming at the service reasoning system and multi task dynamic planning method proposed in this paper,taking the intelligent space of service robot laboratory of Shandong University as the experimental platform,the simulated experiments are carried out to verify the effectiveness and feasibility of the method proposed in this thesis.