Substituting and Augmenting Humans with Robots

Humans have attempted to develop substitutes able to execute their tasks and mimic their behavior. Robots represent the answer to this research, which leads to industrial robots, social humanoids robots, prostheses, that can replace humans in difficult tasks, help them like a relative and restore their lost limbs. Moreover recent years have seen the growth of robots to augment and enhance healthy humans, providing supernumerary robotic limbs that open a range of new possibilities. This thesis aims to investigate the use of robots to both substitute and augment humans. The first part presents three substitution scenarios: i) teleoperation of a 7 dof robotic manipulator using M-IMU sensors; ii) teleoperation of a virtual avatar using stereotaxic systems to investigate multisensory integration; iii) development of a robot-aided Transcranial Magnetic Stimulation (TMS) platform to substitute and improve the performance of expert operators. The second part focuses indeed on the use of supernumerary limbs, addressing three main aspects: i) how humans perform in three-hands tasks; ii) control of a supernumerary limb through high density EMG interface; iii) a vibrotactile interface to provide proprioceptive feedback of a supernumerary robotic limb. Besides proposing useful a reproducible implementation, this thesis also presents the validation of the proposed solutions, whose main results can be resumed as follows: i) in teleoperation reaching tasks with an anthropomorphic robotic arm constraining the robot elbow configuration to be as close as possible to the human elbow one improves the control performance; ii) the proposed VR platform allows to investigate multisensory integration and confirmed the existence of an-hand-centered peripersonal space; iii) the proposed robotaided TMS platform reresents a low-cost and reproducible alternative to the few devices commercially available; iv) the optimization approach allows to significantly decrease the calibration errors regardless of the mathematical implementation; v) the platform allows to elicit MEP amplitudes comparable with the ones measured in manual sessions, reducing the coil orientation error by 46% with respect to expert operators; vi) in three hands tasks people perform better in 3-coupled task than in independent or 2-coupled and in general perform better in couple rather than in solo sessions; vii) the validation of the proposed EMG interface for 2 dof control proved the system proper functioning and easiness of use; viii) the proposed vibrotactile interface allows to effectively identify the end-effector position of a supernumerary robotic limb when the proprioceptive feedback is conveyed as end-effector cartesian information.