Biomechanics and tactile sensing of the human hand : measurement, modelling, and application

The human hand is a marvel of high dexterity with a sophisticated musculoskeletal structure enabling skilled manipulations and object perception. A deep understanding of its associated biomechanics and tactile sensing mechanisms remains at a primitive level, despite their critical importance for the development of robotic or prosthetic hands. So far, primary studies on hand biomechanics, tactile sensing, and sensorimotor mechanisms have been applied to design and control robotic hands. The overall aim of this project is to define a new and effective method for in-depth studies of the human hand biomechanics and tactile sensing, by overcoming the difficulties of high-demand in-vivo measurements. Toward these goals, the following contributions have been made: (1) creation of the first subject-specific anatomically intact finite-element (FE) hand model; (2) quantification of the extensor mechanism and flexible finer joint effects on the hand grasping quality; (3) creation of a novel integrated numerical model predicting afferent tactile signals and exploring the human tactile sensing mechanism; (4) representation of the human sensorimotor mechanism as a transduction function and its application to an in-house soft robotic system with neuromorphic tactile feedback. A subject-specific FE human hand model was created based on the medical images including the entire hand skeleton, subcutaneous tissue, dermis, and epidermis. In-vivo grasping tests were then performed. Muscle forces and hand kinematics were captured to define the loading and boundary conditions. Simulation results were compared to experimental measurements for validation. The FE hand model can accurately represent the biomechanics and performance of the human hand. It was found that the inter-connected extensor tendon and flexible finger joints are the key anatomical features to maintaining the grasping quality and dexterity of the human hand. To study the human tactile sensing mechanism, a novel numerical system integrating an FE human hand model with a neural dynamic model to predict the cutaneous neural response was created. In contrast to other numerical models, which can only compute the afferent tactile signals under passive stimuli, the integrated model proposed here can effectively predict the group cutaneous response during active touch. A microneurography test was conducted and applied to optimise the neural dynamic model. Group responses from cutaneous neurones during active touch were first computed and related to the resulting perception. Temporal 15 coding may be used for the rapid identification of stimuli and triggering reactions, whereas rate coding can represent the quantities of the stimulus. The dynamic relationship between the afferent tactile input and output efferent motor signals can be represented as transduction functions and applied to the control of the robotic/prosthetic hand. A custom-made artificial tactile sensory system was created to implement the accuracy and implications of sensorimotor functions in robotics/prosthetics. A fully 3D printed tactile sensor array was fabricated and mounted on a tendon-driven biomimetic hand. A neural dynamics model was integrated with a tactile sensor to produce neuromorphic tactile signals. The sensorimotor function was applied to control the biomimetic hand and perform different grasps based on neuromorphic tactile feedback. Similar sensing capability and hand performance were achieved by the artificial tactile sensory system compared to human subjects. This constitutes a step further toward the application of next-generation neuroprosthetics.