Tactile Sensing System for Biomechatronic Prosthetic Hands and Grasp Analysis

The development of the artificial tactile sensing began in the 1970s. During the last decades, tactile sensors based on different sensing principles have been developed; the emulation of the physiological sense of touch has gained growing interest in fields such as robotics, as the modern robots are increasingly moving towards unstructured environments. Thus, the acquisition of information about interaction forces and surface properties (such as roughness, hardness, shape etc.) became highly desirable. Several technological solutions have been employed to design tactile sensors; nonetheless, their inclusion within artificial systems still remains problematic in several fields. This is also true when the prosthetic domain is considered. Although tactile sensing is an essential element for autonomous dexterous manipulation, no prosthesis endowed with tactile sensors is available in the market. The present PhD thesis deals with the sensorization of a mechatronic, prosthetic hand, featuring the twofold objective of: i) allowing the control system embedded in such a prosthesis to use sensory information in a closed-loop control, optimizing the applied forces (e.g., during a grasping task) and preventing the manipulated object from slipping; ii) providing the prosthesis user with the aforementioned sensory information, obtaining a closed-loop system which gives the amputee the possibility to properly interact with the external world through the tactile feedback. Starting from the study of human and robotic hands, particularly in terms of tactile characteristics, a solution for an artificial sensory system adoptable for prosthetic hands has been elaborated. This system has been evaluated on a mechatronic hand, namely IH2 Azzurra (Prensilia srl, Pontedera, Italy), similar to the commercial ones but conceived for research scenarios and thus open to both hardware and software modifications. In order to understand the pressure applied onto the objects and its hardness level, force regulation is crucial as well as the avoidance of slippage events. To this end, force and slip sensors are needed for an effective prosthesis sensorization. With the aim of deepening the comprehension of the aforementioned applied pressure, a preliminary evaluation of the grasping forces exerted by human subjects has been done by means of instrumented objects, purposely designed (through CAD software) and developed. These objects have common shapes and dimensions as objects of daily life, but are equipped with sensors for monitoring various physical quantities (such as force, acceleration, temperature etc.). The developed objects have been provided with piezo-resistive sensors for force measurement and with an accelerometer for estimating object orientation and detecting slippage. Suitability for the performance of some Activities of Daily Living (ADLs), such as bi-digital and tri-digital grasps, has driven the design of the objects. The quantities measured with the objects can help obtain insight into the grasping strategies employed by the users. The objective of measuring the force levels applied by an artificial hand can be accomplished by means of tactile sensors. This leads to the sensorization of the above mentioned prosthetic hand. First, a given transduction technology has been chosen within the available ones for the force measurement. This choice has been followed by the identification of the areas to sensorize, which have coincided with the five fingertips, with the metacarpophalangeal joints (MCP) and with the thenar eminence. These areas have been redesigned through CAD software for the inclusion of tactile sensors, given that the IH2 Azzurra (hereafter indicated as IH2) is not equipped with this kind of sensors. Piezo-resistive sensors have been employed: opportune circuitry has been developed for the conversion of the sensors output (i.e., electrical resistance variation) into a force signal, with the realization of dedicated PCBs for the sensors output collection and conditioning. Moreover, an algorithm for identification of slip phenomena has been developed. This algorithm can be easily applied on the force sensor output, allowing the measurement of the force and the generation of an ON/OFF signal relating to the presence or absence of slip events during grasp. The algorithm has rigorously been validated on different typologies of resistive sensors and used for an online implementation into a prosthetic control.