Research On Motion Control Of Human-Robot Interaction For Upper Limb Rehabilitation Robot

With the robot technology have applied to the field of stroke medical rehabilitation more and more popular,many aspects of rehabilitation robot researches has made considerable progress and development,the quality of robot's motion control algorithm affects the user's experience directly.The rehabilitation robot is a "human-robot interaction" system that integrating mechanical technology,embedded technology,sensor technology,control algorithm and rehabilitation medicine.Upper limb rehabilitation robots need robots and human arm to coordinate in harmonic motion,human-robot interaction(HRI)motion works on compliance control algorithm helps to make the performances(maneuverability)satisfactory.We analyzed the existing problems of the prototype according to experience of participating in the upper limb rehabilitation robot project.Among the rehabilitation robot prototype's developing process,the improvement of machinery and electrical makes the performance of the robot has improved overall,but the core control algorithm of the rehabilitation robot has some defects,the intuitive performance in the human-robot interaction process is that human-robot interaction(HRI)coordination movement is not good,and the catastrophe and other intuitive phenomena need to find the fundamental problem,and then integrated in the control algorithm to solve the problem.In order to study the effect of variety single impact on the interactive performance,the multidimensional rehabilitation robot system is simplified.A one degree of freedom(DOF)"hardwarein-the-loop" experiment system is built.Based on this system,the following studies were carried out:The data of force sensor signal which is used to characterize the change of mechanical properties in the process of human-robot interaction(HRI)needs to be collected in real time and then feed back to the control system.(1)Six-axis high-precision force / torque(F / T)sensor pre-signal processing.The data of axis force sensor signal which is used to characterize the change of mechanical properties in the process of human-robot interaction(HRI)needs to be collected in real time in order to feed back to the control system.Taking about that the weak transmission sensor signals will be interfered by random noise during interaction such as electromagnetic interference,the accuracy and accuracy of the control system will be affected,the sensor signal need for pre-filtering.Based on a variety of digital filtering algorithms,Kalman filter is used to process the collected signals,and the ideal effect is achieved.(2)Complex friction identification.In the electromechanical system,the friction force exists objectively.As the rehabilitation robot equipment of the human-robot interaction system,the manual experience is not only affected by the friction of the secondary surface of the motion,but also the use of the harmonic reducer makes the negative performance of the friction.In this paper,the characteristics of various dry friction models are analyzed.Finally,LuGre friction model is used to identify the friction of the system.It is proved that the friction model can accurately characterize the friction characteristics of the system.The influence of LuGre friction model on the system is studied.The experimental results verify the rationality of the model selection and parameter identification.(3)Optimize the system control strategy.In this paper,the accurate identification friction is taken as feedforward compensation,the characteristics of the algorithm are simulated and experimentally studied.The results show that the model based friction compensation impedance control shows a better compliance control effect.(4)In order to improve the flexibility of human-robot interaction,the paper studied the control strategy based on the haptic characteristics,and is going to study the impact of human-robot interaction flexibility factors from many aspects.