Study On A Variable Stiffness Robotic Joint And Human-Robot Joint Stiffness Matching

With the development of technology,robots are increasingly used in life services,medical assistance,exoskeleton and other fields,which require higher demand for human-robot interactions.Traditional rigid robots cannot guarantee the security of human-robot physical interactions.Therefore improving human-robot compliant interaction becomes an urgent problem in the field of cooperative robots.Human use muscles to drive their joints.Muscles can change their stiffness actively so as to enable better interactive abilities.Based on the functioning principle of human muscle system,this paper carried out the research on compliant robot design and control technology.A soft robotic joint with variable stiffness was designed through mimicking human elbow.Surface electromyography signals(sEMG)were used to represent human elbow stiffness.A compliant control strategy for the robotic joint was proposed to improve the performance of human-robot interactions.The main research contents of this paper are as follows:Based on the bionic principle of human elbow joint,a soft variable-stiffness robotic joint SoftJoint I was designed.The robotic joint was driven by a motor and a rotary SEA(Serial Elastic Actuator).The SEA is composed of linear springs arranged circumferentially.Mimicking antagonistic joint muscles,a group of elastic bands arranged in symmetry with respect to the joint axis are employed to achieve variable stiffness of the robotic joint.Elastic bands were controlled using an stiffness adjusting motor.SoftJoint I prototype is built.Based on static analysis,the stiffness model of the robotic joint was derived.The joint stiffness is the function of the elastic bands length,the joint angular displacement and the joint deformation angle.The joint stiffness characteristic were tested which verify the correctness of the joint stiffness mathematical model.The testing results showed that the robotic joint has good stiffness response characteristics and dynamic motion performances.sEMG signal based method was employed to determine the stiffness of human elbow joint.sEMG signals were collected using MYO armband.The sEMG signals were processed with amplitude envelope and normalization.Based on the active level of human muscles,sEMG amplitude was segmented and mapped to be the human elbow joint stiffness.The experimental platform of human-robot collaboration was built,which composes SoftJoint I prototype and its control system,MYO armband and sEMG signal collecting system.The experimental platform uses a upper and a lower computers to execute the functions of joint motion drive,stiffness adjustment,sEMG collecting and processing,human joint stiffness mapping and human-robot interaction control strategy.According to human arm interactive characteristics,a compliant control strategy based on human and robot joint stiffness matching was proposed.During human-robot collaborations,the human joint stiffness evaluated using sEMG signals was employed as the desired joint stiffness of SoftJoint I.SoftJoint I controls the stiffness adjusting motor to match the desired joint stiffness for better cooperative performances.The proposed compliant control strategy improves the effectiveness of human-robot collaboration.The robotic joint achieves the ability to adjust its stiffness according to task changes and to adapt to the changing environments.