Multiple Hand Gesture Recognition using Surface EMG Signals

Abstract

Significance of robotics in serving the human being is increasing day by day. A large number of impairments and disabilities in human body force the researchers to think on the necessity of simple and natural human-machine control interface. The idea of the project is the acquisition of SEMG (Surface Electromyographic) signals from the forearm and to recognize the various hand gestures. The resulting classification is then used to control a two degree of freedom (DOF) robotic gripper. Muscular activity is sensed by placing the EMG sensors/electrodes on the skin. The acquired signal from these electrodes is very small in amplitude and corrupted by different artifacts due to positioning and pasting of electrodes, transmission line and crosstalk with other biological signals. Pre-amplification is required to boost up the signal and then filtration is required to get the desired usable band of frequency. After that artifact-free EMG signal is further amplified, which can be fed to the control circuitry (microcontroller) to control the Hobby Servo motor of the robotic gripper hand. All the process is implemented for the real time scenario.