Neuroscientific and soft-robotic principles for a new generation of natural bionic limbs.

Compensating for losing the fine and coordinated functions of human upper extremities with prostheses is a medical, technological, psychological, and social challenge. Even though existing artificial limbs promise to restore some of those missing capabilities, there is still a wide gap between what available commercial devices offer and what users demand. While most commercial prostheses present rigid mechanical structures, emerging trends in the design of robotic hands are moving towards the introduction of soft technologies, capable to adapt their behavior according to changes in their configuration, in the environment, or according to external circumstances. Although this promising approach is inspired by nature and could be innovative for prosthetic applications, there is scant literature concerning its benefits for the end-users. The first contribution of this thesis is the evaluation of diverse stiffness levels and variable stiffness controllers in able-bodied subjects to understand the role of stiffness in humans. Then, I investigate the role and the benefits of soft robotic technologies in the field of prosthetics. Specifically, in collaboration with Usl Toscana NordOvest (Italy), I studied its effect on functionality, embodiment, and user-perception through different clinical assessment tools on unilateral and bilateral individuals with limb loss. The results of those studies, which agree with some of the intuitions found in the literature and our experience in Cybathlon, highlight important occasions for improvement. Some of these are studied further in my thesis, which develops more advanced biomechanics and algorithms for the creation of a new generation of prostheses. To implement those insights into practical solutions, my thesis focuses on the dual aspects of softness and synergies in human motor control. Postural synergies are considered guiding and simplifying the design, whereas muscular and neuronal synergies led to improvements in the intuitiveness and functionality of myoelectric control. One important class of improvements concerns advances in the mechanics of the hand and the wrist, where dexterity should be provided to users while preserving simplicity and usability. Here, I introduce novel solutions that leverage the principles of under-actuation, aiming to reduce compensatory movements. My work presents insights regarding the introduction of an articulated palm to increase the contribution of all hand parts in grasping, the use of dynamic synergies in prostheses, and multisynergistic hands for the continuous exploration of grasping patterns and in-hand manipulation. For the wrist, I propose a compact 3 DoF myoelectric semi-passive wrist capable of locking in different positions and able to adapt its mechanical impedance, from compliant to rigid, to favor natural body postures. Finally, my work explores different control strategies to introduce adaptability regulation in the human-robot interface and improve the sharing of intelligence toward natural methods. I explored the effect of voluntary and involuntary impedance control of prostheses in performing Activities of Daily Living and social interactions. Then, in collaboration with Prof. Dario Farina at Imperial College London, I proposed alternative myoelectric control strategies based on the functional organization of motor units and neuronal synergistic organization information, especially useful for multisynergistic hands.