Trust Sensor Interface for Improving Reliability of EMG-based User Intent Recognition

View Research Poster

The Problem

Electromyographic (EMG) signal, which is widely investigated for natural and smooth control of prosthesis, can be easily contaminated by diverse disturbances. The disturbed EMG signal may lead to errors in user intent identification (UII), and even cause dangers (e.g. tumbles and falls) to the amputees.


To address this problem, we proposed the trust sensor interface (TSI). TSI contains 2 modules: the abnormal detector, which detects diverse disturbances with high accuracy and low latency, and the trust evaluation, which dynamically evaluates the reliability of EMG sensors. Based on the output of TSI, UII is able to dynamically adjust its operations or decisions (i.e. sensor add-in or removal, system re-calibration) when disturbances occur, which improves the reliability of the output of UII, and therefore improves the safety of amputees.


Our experiment on a human subject demonstrated the feasibility of the TSI and improvement in the reliability of TSI. The video demonstration can be found on YouTube.


This research is partially supported by NSF Award # 0931820.