Research On Human Robot Interaction And Motion Control Of Mobile Robot

With the gradual popularization of intelligent service robots,a more natural human-robot interaction has become a necessary function of the robot.In the way of communication between people and people,"talking",which is based on the interaction between speech and natural language,is undoubtedly the most natural,direct and simple interactive way.Therefore,an intelligent service robot should also use this way to realize the interaction between man and robot.In this paper,a robot human-robot interaction and control system based on speech and natural language is designed to solve the human-robot interaction problem of intelligent service robot.First,according to the functional requirements of robot human-robot interaction and control system,the system is divided into three subsystems,namely,speech human-robot interaction system,natural language instruction understanding system and robot control system,and subsystems are designed separately.Secondly,aiming at the core problem of human-robot interaction system,that is,robot's natural language instruction understanding,a robot natural language instruction understanding method based on instruction intention understanding and key information extraction is proposed.The method mainly includes two parts,the instruction intention understanding based on the short text classification and the key information extraction based on the sequence annotation.The former transforms the instruction intent understanding problem into short text classification problem,and realizes the instruction intention understanding based on machine learning and deep learning short text classification model in Natural Language Processing.The latter transforms the key information extraction problem to the sequence annotation problem,and realizes the instruction key information extraction by using the sequence annotation method and the conditional random field model in Natural Language Processing domain.At the same time,a lot of natural language command corpus is collected for the family service robot environment,and it is analyzed and processed as a training corpus for instruction understanding model.Finally,the ROS robot operating system and the cloud server were used to build the human-robot interaction and control system of the intelligent service robot.The speech human-robot interaction system includes the voice interaction in the far field and the WeChat voice interaction.The natural language instruction understanding system is deployed on the cloud server side and invoked through the network.The robot control system implements a distributed,multi-level and modular robot control system through ROS,and takes the motion control task module as an example to elaborate the implementation process.The three part of the organic integration,constitutes a complete set of intelligent service robot human-robot interaction and control system.