Abstract:
People with paralyzed upper limbs and children with cerebral palsy do not have a controlled movement.
The messages from their brain do not transmit properly and the movement is either completely stopped or
uncontrolled. Electroencephalogram (EEG) signals and brain-computer interface (BCI) have a great
impact on assistive and rehabilitation gadgets. This project focuses on an exoskeleton, which takes EEG
signals from the brain and moves the exoskeleton using BCI. The mechanical system consists of one active
degrees of freedom of elbow, and one passive support joint of the shoulder. Shoulder and wrist joints will
be passive joints and elbow joint is an active joint. The system can be used for daily tasks, weight bearing,
and therapies. We use an open-access two class EEG data set for the processing. The data was filtered,
and artifacts were removed. After preprocessing, the feature extraction technique CSP, namely Common
Spatial Pattern, was used and the data was classified using four different classification algorithms, out of
which, Quadratic Discriminant gave the best results.
