Electronic systems for the restoration of the sense of touch in upper limb prosthetics

In the last few years, research on active prosthetics for upper limbs focused on improving the human functionalities and the control. New methods have been proposed for measuring the user muscle activity and translating it into the prosthesis control commands. Developing the feed-forward interface so that the prosthesis better follows the intention of the user is an important step towards improving the quality of life of people with limb amputation. However, prosthesis users can neither feel if something or someone is touching them over the prosthesis and nor perceive the temperature or roughness of objects. Prosthesis users are helped by looking at an object, but they cannot detect anything otherwise. Their sight gives them most information. Therefore, to foster the prosthesis embodiment and utility, it is necessary to have a prosthetic system that not only responds to the control signals provided by the user, but also transmits back to the user the information about the current state of the prosthesis. This thesis presents an electronic skin system to close the loop in prostheses towards the restoration of the sense of touch in prosthesis users. The proposed electronic skin system inlcudes an advanced distributed sensing (electronic skin), a system for (i) signal conditioning, (ii) data acquisition, and (iii) data processing, and a stimulation system. The idea is to integrate all these components into a myoelectric prosthesis. Embedding the electronic system and the sensing materials is a critical issue on the way of development of new prostheses. In particular, processing the data, originated from the electronic skin, into low- or high-level information is the key issue to be addressed by the embedded electronic system. Recently, it has been proved that the Machine Learning is a promising approach in processing tactile sensors information. Many studies have been shown the Machine Learning eectiveness in the classication of input touch modalities.More specically, this thesis is focused on the stimulation system, allowing the communication of a mechanical interaction from the electronic skin to prosthesis users, and the dedicated implementation of algorithms for processing tactile data originating from the electronic skin. On system level, the thesis provides design of the experimental setup, experimental protocol, and of algorithms to process tactile data. On architectural level, the thesis proposes a design ow for the implementation of digital circuits for both FPGA and integrated circuits, and techniques for the power management of embedded systems for Machine Learning algorithms.