Abstract
This paper describes a process for assessing multi-loop wearable devices that use a common slider to passively drive the exo-fingers for the physical training of people with limited hand mobility. Each finger design, exceptfor the thumb, is based on an RRR serial chain, termed backbone, constrained into a multi-loop eight-bar slider mechanism using two RR constraints. The thumb utilizes a planar RR backbone chain constrained into a parallel four bar slider. During the physical task acquisition experiments, the subject's tip finger trajectories are captured using an optical motion capture and its dimensions are set such that they match each of the fingers kinematics as closely as possible. The dimensional synthesis procedure can yield a variety of design candidates that fulfill the desired fingertip precision grasping trajectory. Once it is ensured that the synthesized fingertip motion is close to the physiological fingertip grasping trajectories, performance assessment criteria related to user-device interference and natural joint angle movement are taken into account. After the most preferred design for each finger is chosen, minor modifications related to substituting the backbone chain with the wearer's limb to provide the skeletal structure of the customized passive device are made. To illustrate the proposed technique, the development of a 3D prototype model of a passively actuated Closed Loop Articulated Wearable (CLAW) hand is presented. The CLAW hand performance with respect to wearability and robustness was assessed. Preliminary test results with healthy subjects show that the CLAW hand is easy to operate and able to guide the user's fingers without causing any discomfort, ensuring both, precision and power grasping in a natural manner. The lack of electrical actuators and sensors simplifies the control, resulting in a lightweight and cost-effective solution for grasping of a variety of objects with different sizes. This work establishes the importance of incorporating novel design candidate assessment techniques, based on human finger kinematic models, within the conceptual design level that can assist in finding robust design candidates with naturalistic joint motion.