The fuzzy controller above utilizes the user surface EMG activities to move the elbow joint of the robotic arm. The user uses multiple biceps contractions to accurately control the elbow joint position. The controller accepts as inputs the normalized surface EMG amplitude and its slope as inputs. The output of the fuzzy controller is the accurate direction and magnitude of the fuzzy controller. The filtering algorithm utilizes the Mean Absolute Value (MAV) of the raw EMG signal. When the user repeatedly flexes her bicep, the robotic arm elbow joint flexes in the clockwise direction, when the user relaxes her biceps the robotic arm elbow joint moves in the counter-clockwise direction. The accuracy of the biceps contraction and the direction of the robotic arm elbow joint is determined by the fuzzy controller. Also, the fuzzy controller makes the system immune to crosstalk.

Myoelectric Control using Linear Discriminant Analysis with Autoregressive Features

The video displays a control scheme for a robotic arm. The input of the controller is the output of a classifier. The classifier used is a linear discriminant analysis which has an output of either flexion, extension, or rest. The input to the classifier comes from autoregressive features extracted from biceps and triceps EMG signals.