Real-Time Contact Sensory Feedback for Upper Limb Robotic Prostheses

Abstract

Lack of sense of touch is a fundamental problem for today's robotic prostheses. Considering that touch feedback plays a significant role in identifying contacted objects, the aim of this study is to use acceleration signals, which occur due to physical contact of a prosthetic hand with objects, as sensory feedback. These contact signals are applied on the clavicle bone using a tactor. First, a library of accelerations occurring as a result of tapping on different materials is prepared. The effect of impact speed is studied and used as a scaler for real-time applications. A stochastic signal model is developed for the contact accelerations. Due to the distinct waveform characteristics of different materials, the rate of change of acceleration is used to identify hardness of the objects. In a psychophysical study, the whole procedure of recording, identifying, and replaying the signals is investigated. Results of this study showed the ability of the designed tactor to provide distinguishable hardness sensations of different materials in real time. Furthermore, considering the results of the position control test with and without vision, it can be concluded that the proposed contact sensory feedback method could be enough to substitute for vision during simple grasping tasks.