Human-operated And Human-Like Adaptive Impedance Controls On Exoskeleton Robot

To transfer the skilful motion dexterity of humans to the robot manipulation,It is necessary to collect the physical signals(i.e. velocity, acceleration), and the physiology signals(i.e. surface electromyography(sEMG) signals) as well. It is well known that humans are able to adapt limb force and impedance in various interactive scenarios, e. g., tool using and object manipulation. However, most tele-operation robots based on force control do not have physical elastic properties. In addition, to acquiring the elastic properties, many improved tele-operation interfaces brought down the transparency of system and restricted the interaction between human and robot. sEMG based technique can increase the transparency and have the advantage that it does not depend on the precious hardware. Based on the relevant researches, we utilized the sEMG signals to establish a novel human machine interface. In detail, the impedance information is acquired from the sEMG signals of the human limbs and then mapped to joint space and Cartesian space of exoskeleton robot. The human skill transfer is performed by the torque-based controller. This novel interface could realize a stable and safe human machine interaction.In this paper, two methods are proposed to investigate the impedance profiles from the sEMG signals. First of all, a straightforward model which assumes a linear relationship between end-point stiffness and rectified sEMG signals was applied in the fixed posture configuration. On the other hand, by exploring the configuration-dependent properties of the joint and Cartesian stiffness, a reduced- complexity model of arm endpoint stiffness was applied to estimate the Cartesian stiffness from the sEMG signals. In this work, we combined human transferred impedance into the relevant control methods(disturbance observer and the robust adaptive control). The effectiveness of the human skill transfer is testified in different motion conditions(stational station and moving dynamic station). Finally, to verify the performance of the proposed method, a 19 degree of freedom exoskeleton robot was used in the related experiments while the human transferred impedance is inserted into the control of the real-time manipulation of exoskeleton robot. The result demonstrated that the advanced controller with human transferred stiffness performed a better property in disturbance rejection, when compared with the traditional control method. The proposed control framework may offer helps in passive rehabilitation.