Brain Machine Interface Controlled Robot Chair

• Main Author: Hema Chengalvarayan, Radhakrishnamurthy
• Format: Thesis
• Language: English
• Published: Universiti Malaysia Perlis (UniMAP) 2010
• Subjects: Controlled robot chair; Brain machine interface (BMI); Electroencephalography (EEG); Wheel chair
• Online Access: http://dspace.unimap.edu.my/xmlui/handle/123456789/9860

Brain Machine Interface Controlled Robot Chair: Brain Machine Interface is a device that links the human brain directly to devices such as computer, wheelchairs and prosthetic arms. Such interfaces provide a digital channel for communication and control in the absence of the biological channels and thus help in the rehabilitation of mobility and speech impaired individuals. In this thesis, a novel four-class brain machine interface (BMI) is designed for a robot chair using neural networks. Simple and novel protocols for acquiring brain EEG signals from two non-invasive scalp electrodes are presented. Four tasks based on motor imagery of left and right hand movements are proposed to control the directions of the robot chair. A novel algorithm for acquisition of motor imagery signals using only hand movements is proposed. Simple preprocessing algorithms are presented to remove noise from the raw signals. Mu, Beta and Gamma frequency bands related to the motor actions are extracted using customised filters. New features based on time and frequency components of the EEG signals are proposed and tested with classifiers. Classification of the four hand motor imagery signals is presented using static and dynamic neural networks. A particle swarm optimization based algorithm is proposed to train the neural networks. Combinations of the features proposed and the static and dynamic classifiers are analysed. Signals collected from 10 trained subjects are used in the analysis of synchronous and asynchronous BMI designs. A max-one algorithm for translation of the hand motor imagery signals into robot chair movements is presented. A prototype robot chair is designed and interfaced with the developed asynchronous BMI. Safety features are integrated through a collision avoidance system to enhance the performance of the robot chair. The BMI controls the joystick of the robot chair using a shared control algorithm. Real-time experiments are also presented using 10 trained and 5 untrained subjects to validate the applicability of the brain machine interface. Experiments were carried out at two expositions (out-of-lab environments) with 25 untrained subjects to assess its feasibility in real life environments.