Font Size: a A A

The Research Of Human-robot Interaction Strategies And Intelligence Learning Based On Teleoperated Robotics

Posted on:2021-05-06Degree:DoctorType:Dissertation
Country:ChinaCandidate:J LuoFull Text:PDF
GTID:1368330611967148Subject:Pattern Recognition and Intelligent Systems
Abstract/Summary:PDF Full Text Request
With the rapid development of control,sensor,and machinery,robotics are widely used in our production and life with good application prospects.As a branch of robotics,teleoperated robot has attracted much attention from the academia and industry due to its own capabilities of teleoperation and hazardous operation.However,restricted by many factors,e.g.,poor human robot interaction,weak perception and learning,etc.,applications of the teleoperated robots are limited.In this sense,how to deal with these problems are becoming the key factors for the development of teleoperated robots with important theoretical values and high practical application values.This work proposes a human-robot interaction control and intelligence learning method to achieve a safer and more natural interaction between the human and the robots from the human aspect for the teleoperated robotic system.The proposed method is used to improve intelligence of the teleoperated robotic system as much as possible and mainly describes the relationship of human factor and the teleoperated robots.It is an intersecting subject of control science,computer science,mechanical science,artificial intelligence technology,and biological signal processing technology for the technique of human-robot interaction control and intelligence learning of the teleoperated robotic system.It studies on the human-robot interaction control,the robot task learning model,the robot's perception learning,etc.,and combines the robot's advantages with the human guidance and intelligence to achieve co-operation.The main contributions and innovations can be concluded as follows.(1)In order to achieve a safe and friendly human-robot interaction control performance of the teleoperated robots,this paper proposes a control scheme based on operator's manipulation characteristics.This paper analyzes the muscle activation in the process of teleoperation and proposes a method which combining variable gain control and hybrid control for the purpose of friendly human-robot interaction.At the same time,a tremor filter based on support vector machine is developed to eliminate the influence of the tremor of the operator.In the end,the effectiveness of the proposed human-robot interaction control scheme is verified by the comparative experiments.(2)To enhance the performance of human-like manipulation and perception learning,an integrated framework based on robot task learning and human-robot interface is developed in this thesis.The relationship among human operator,teleoperated robots,and the task is explored to model the human-robot interaction task for generating a continuous and smooth task model and obtaining the manipulation skill of human operator.The framework can naturally percept the feedback of interaction environment through the variety of muscle activation and haptic perception of teleoperated robots,and send the correct control commands to the remote robots.In addition,the proposed method can obtain the task model and control intention of human operator through characterization of interaction task trajectories and muscle stiffness of human operator.Finally,the different experiments demonstrate the feasibility of the proposed humanlike manipulation and perception learning method.(3)A teleoperation control scheme based on hybrid shared control is proposed to enhance the performance of obstacle avoidance for human-robot interaction.Based on the artificial potential field method,the hybrid shared control with muscle activation can deal with the problems of obstacle avoidance and interaction.Finally,the experimental results demonstrate the effectiveness and superiority of the proposed method.
Keywords/Search Tags:Teleoperated robots, human-robot interaction control, task learning and perception, control interaction, hybrid shared control
PDF Full Text Request
Related items