Analysis of Electroencephalographic Signal Acquisition and Processing for Use in Robotic Arm Movement

In recent years, new research has brought the field of electroencephalograph (EEG)-based brain– computer interfacing (BCI) out of its early stages into a phase of relative maturity through many demonstrated prototypes to assist the periphery compromised patients. It is a worth- while technology for disabled people enabling them to reinstate a damaged motor nerve or any neural pathway. This study introduces the development of Brain computer interface by means of a non-invasive wireless electroencephalograph (Emotiv EPOC system) and its application to control a prototype robotic arm. After the pre-processing and spectral analysis of raw EEG signals, an averaged peak value for selected channels were obtained as a feature vector for each movement. Two algorithms namely, linear discriminant analysis (LDA) and quadratic discriminant analysis (QDA) were used to differentiate the raw EEG data into their associative movements. Performance of a classification algorithms were assessed, which revealed that the accuracy was 71.77 (±0.76) % and 86.57(±0.79) % of LDA and QDA respectively. Feature vector resulted in superior performance of 86.57 (±0.79) % with QDA. The averaged peak value of PSD for selected seven channels were then used to move a robotic arm successfully in the two directions i . e . "Up (elbow flexion)" and "Down (elbow extension)".