Abstract
The integration of human and robot into a single system offers remarkable opportunities for a new generation of assistive technology. Despite the recent prominence of upper limb exoskeletons in assistive applications, the human arm kinematics and dynamics are usually described in single or multiple arm movements that are not associated with any concrete activity of daily living (ADL). Moreover, the design of an exoskeleton, which is physically linked to the human body, must have a workspace that matches as close as possible with the workspace of the human body, while at the same time avoid singular configurations of the exoskeleton within the human workspace. The aims of the research reported in this manuscript are (1) to study the kinematics and the dynamics of the human arm during daily activities in a free and unconstrained environment, (2) to study the manipulability (isotropy) of a 7-degree-of-freedom (DOF)-powered exoskeleton arm given the kinematics and the dynamics of the human arm in ADLs. Kinematic data of the upper limb were acquired with a motion capture system while performing 24 daily activities from six subjects. Utilising a 7-DOF model of the human arm, the equations of motion were used to calculate joint torques from measured kinematics. In addition, the exoskeleton isotropy was calculated and mapped with respect to the spacial distribution of the human arm configurations during the 24 daily activities. The results indicate that the kinematic joint distributions representing all 24 actions appear normally distributed except for elbow flexion–extension with the emergence of three modal centres. Velocity and acceleration components of joint torque distributions were normally distributed about 0 Nm, whereas gravitational component distributions varied with joint. Additionally, velocity effects were found to contribute only 1/100th of the total joint torque, whereas acceleration components contribute 1/10th of the total torque at the shoulder and elbow, and nearly half of the total torque at the wrist. These results suggest that the majority of human arm joint torques are devoted to supporting the human arm position in space while compensating gravitational loads whereas a minor portion of the joint torques is dedicated to arm motion itself. A unique axial orientation at the base of the exoskeleton allowed the singular configuration of the shoulder joint to be moved towards the boundary of the human arm workspace while supporting 95% of the arm's workspace. At the same time, this orientation allowed the best exoskeleton manipulability at the most commonly used human arm configuration during ADLs. One of the potential implications of these results might be the need to compensate gravitational load during robotic-assistive rehabilitation treatment. Moreover, results of a manipulability analysis of the exoskeleton system indicate that the singular configuration of the exoskeleton system may be moved out of the human arm physiological workspace while maximising the overlap between the human arm and the exoskeleton workspaces. The collected database along with kinematic and dynamic analyses may provide a fundamental basis towards the development of assistive technologies for the human arm.