Design-validation of a hand exoskeleton using musculoskeletal modeling.

Exoskeletons are progressively reaching homes and workplaces, allowing interaction with virtual environments, remote control of robots, or assisting human operators in carrying heavy loads. Their design is however still a challenge as these robots, being mechanically linked to the operators who wear them, have to meet ergonomic constraints besides usual robotic requirements in terms of workspace, speed, or efforts. They have in particular to fit the anthropometry and mobility of their users. This traditionally results in numerous prototypes which are progressively fitted to each individual person. In this paper, we propose instead to validate the design of a hand exoskeleton in a fully digital environment, without the need for a physical prototype. The purpose of this study is thus to examine whether finger kinematics are altered when using a given hand exoskeleton. Therefore, user specific musculoskeletal models were created and driven by a motion capture system to evaluate the fingers' joint kinematics when performing two industrial related tasks. The kinematic chain of the exoskeleton was added to the musculoskeletal models and its compliance with the hand movements was evaluated. Our results show that the proposed exoskeleton design does not influence fingers' joints angles, the coefficient of determination between the model with and without exoskeleton being consistently high (R2¯=0.93) and the nRMSE consistently low (nRMSE¯ = 5.42°). These results are promising and this approach combining musculoskeletal and robotic modeling driven by motion capture data could be a key factor in the ergonomics validation of the design of orthotic devices and exoskeletons prior to manufacturing.

Toward Haptic Communication: Tactile Alphabets Based on Fingertip Skin Stretch.

This paper studies the possibility to convey information using tactile stimulation on fingertips. We designed and evaluated three tactile alphabets which are rendered by stretching the skin of the index's fingertip: (1) a Morse-like alphabet, (2) a symbolic alphabet using two successive dashes, and (3) a display of Roman letters based on the Unistrokes alphabet. All three alphabets (26 letters each) were evaluated through a user study in terms of recognition rate, intuitiveness, and learnability. Participants were able to perceive and recognize the letters with very good results (80-97 percent recognition rates). Taken together, our results pave the way to novel kinds of communication using tactile modality.

Analysis of hand contact areas and interaction capabilities during manipulation and exploration.

Manual human-computer interfaces for virtual reality are designed to allow an operator interacting with a computer simulation as naturally as possible. Dexterous haptic interfaces are the best suited for this goal. They give intuitive and efficient control on the environment with haptic and tactile feedback. This paper is aimed at helping in the choice of the interaction areas to be taken into account in the design of such interfaces. The literature dealing with hand interactions is first reviewed in order to point out the contact areas involved in exploration and manipulation tasks. Their frequencies of use are then extracted from existing recordings. The results are gathered in an original graphical interaction map allowing for a simple visualization of the way the hand is used, and compared with a map of mechanoreceptors densities. Then an interaction tree, mapping the relative amount of actions made available through the use of a given contact area, is built and correlated with the losses of hand function induced by amputations. A rating of some existing haptic interfaces and guidelines for their design are finally achieved to illustrate a possible use of the developed graphical tools.

Self-adjusting, isostatic exoskeleton for the human knee joint.

A knee-joint exoskeleton design that can apply programmable torques to the articulation and that self-adjusts to its physiological movements is described. Self-adjustment means that the articular torque is automatically produced around the rotational axis of the joint. The requirements are first discussed and the conditions under which the system tracks the spatial relative movements of the limbs are given. If these conditions are met, the torque applied to the joint takes into account the possible relative movements of the limbs without introducing constraints. A prototype was built to demonstrate the applicability of these principles and preliminary tests were carried out to validate the design.