Passivity Based Adaptive Control for Upper Extremity Assist Exoskeleton

Abstract

Upper limb assist exoskeleton robot requires quantitative techniques to assess human motor function and generate command signal for robots to act in compliance with human motion. To asses human motor function, we present Desired Motion Intention (DMI) estimation algorithm using Muscle Circumference Sensor (MCS) and load cells. Here, MCS measures human elbow joint torque using human arm kinematics, biceps/triceps muscle model and physiological cross sectional area of these muscles whereas load cells play a compensatory role for the torque generated by shoulder muscles as these cells measure desire of shoulder muscles to move the arm and not the internal activity of shoulder muscles. Furthermore, damped least square algorithm is used to estimate Desired Motion Intention (DMI) from these torques. To track this estimated DMI, we have used passivity based adaptive control algorithm. This control techniques is particular useful to adapt modeling error of assist exoskeleton robot for different subjects. Proposed methodology is experimentally evaluated on seven degree of freedom upper limb assist exoskeleton. Results show that DMI is well estimated and tracked for assistance by the proposed control algorithm.