Closing the Loop: Brain Processes underlying Control and Embodiment of Supernumerary Robotic Limbs

Embodiment, i.e. the feeling of possessing and being in charge of a body, is a multifaced concept that requires complex estimations, achieved through multisensory integration, and deals with a settled knowledge according to a Bayesian statistical approach. Embodiment is intimately related to the body schema: a sensorimotor representation of the body that guides actions. Thus, motor output and sensory feedback are inscribed in a closed loop, known as sensorimotor loop, which is considered of paramount importance in the genesis of embodiment itself. However, the relative contributions of multisensory cues and motor planning remain to be assessed. This issue is addressed in the present work by combining the Moving Hand Illusion and the Tendon Vibration Illusion paradigms. Results demonstrated that, in order to feel embodiment of an artificial limb, the congruency between two integrated sources of information is required, be it an inter-sensory or a sensorimotor congruency. Furthermore, it is suggested that the efference copy could be exploited as a sensory modality to doublecheck for the embodiment of an artificial limb. Successively, by delving into the theory of Internal Models and showing that it is possible to induce low-level kinematics adjustment through artificial proprioceptive feedback during a lifting task, it is argued that a similar double role can be played by sensory afference as well. Notions and results expressed up to this point are finally framed in the context of human augmentation, and Supernumerary Robotic Limbs (SRL) in particular. The present work addresses the open question of how to establish a bi-directional communication through the addition of a robot-related proprioceptive sensory feedback, in order to close the sensorimotor loop and achieve a better human-robot interaction. Results suggested that position feedback seems to be a valid option for conveying supplementary proprioceptive feedback. Lastly, the present thesis tackles with another open issue in the field of human augmentation: to really benefit from an SRL, the user needs to receive feedbacks from the SRL seamlessly. By investigating the feasibility of a real-time supplementary feedback processing, the present preliminary work suggests that proprioceptive information regarding the SRL, such as the cartesian space position, could be conveyed and understood by the user with high precision and low delay.