Providing contact sensory feedback for upper limb robotic prosthesis

Abstract

Lack of the sense of touch is the fundamental problem of today's robotic prostheses. Considering the fact that touch feedback plays a significant role on identifying contacted objects, our aim in this study is to use acceleration signals, occurring due to physical contact of a prosthetic hand with objects, as sensory feedback. We apply these signals on the clavicle bone using a tactor as a haptic interface. First, a library of the acceleration signals occurring as a result of tapping on different materials is collected. Effect of the impact velocity is studied and used as a scalar for real-time applications. In order to model the contact accelerations, a stochastic signal model is developed. Due to the distinct waveform characteristics of different materials, frequency and amplitude of the acceleration signals, together with the impact velocity, are used to identify stiffness of the objects. In a real-time experiment, performed using a robotic finger, the whole procedure of recording, identifying and replaying the signals by the tactor is studied. The preliminary results show that the tactor is able to provide distinguishable feelings of hard and soft objects to the user.