Toward Natural Bionic Limbs Through Musculoskeletal and Sensorimotor Embodiment

Prosthetic devices are essential assistive technologies that restore motor function and autonomy after limb loss. Over the past decades, research has considerably improved their functionality and technological sophistication. However, current devices remain far from replicating the versatility and natural behavior of the human arm. Most prosthetic systems still rely on rigid actuation and fixed mechanical impedance to restore motor function, at the expense of safe and natural interaction with the environment and adaptive behavior across tasks. These limitations, together with control strategies that remain unintuitive and poorly aligned with the user's sensorimotor organization, hinder the natural integration of artificial limbs. To address these limitations, I develop a prosthetic framework inspired by the human musculoskeletal and sensorimotor systems, in which redundant actuation, physical compliance, controllable stiffness, and multimodal control strategies are jointly designed to support closed-loop integration of the prosthesis within the user's body schema. In this framework, I develop a variable stiffness elbow prosthesis capable of independently regulating posture and stiffness through redundant elastic actuation. I extend this concept to multi-\gls{DoF} articulations through a 3-\gls{DoF} prosthetic wrist with continuously adjustable stiffness, based on a hybrid serial-parallel architecture inspired by human muscle cocontraction. To address distal interaction with the environment, I develop a soft robotic hand that introduces a novel implementation of adaptive synergies within a modular platform, unifying multiple synergy configurations and application-driven customizations within a single underlying architecture. To control such systems, I propose a novel approach leveraging muscular activity and compensatory body movements as complementary control channels, integrating prosthesis operation within the user's natural motor behavior in a closed-loop fashion. All these technologies are integrated into V-Soft Pro, a modular prosthetic platform that embeds user-controllable stiffness within its design and supports closed-loop integration with the user's sensorimotor system. I validate these developments through experimental studies on individual modules and on the integrated V-Soft Pro system, culminating in a longitudinal pilot study assessing the functional implications of soft robotic prosthetic technologies during home use with the SoftHand Pro. Overall, this thesis demonstrates that integrating musculoskeletal and sensorimotor principles into upper limb prostheses can enhance interaction capabilities, promote more natural prosthesis behavior, and support closer integration of the artificial limb within the user's body schema, establishing these principles as key foundations for next-generation bionic limbs.