Implementation of Artificial Neural Networks for Grasping Activity Detection

A grasping detection system is essential to develop a prosthesis for helping people with disabilities. This study aims to detect grasping activity using an Artificial Neural Network using an Electromyograph sensor. The tool consists of three components: an Electromyograph sensor, an Arduino Nano, and an LED. The Electromyograph sensor detects muscle activity on the lower left arm. There are 5 healthy subject is involved in this study. Data from the subject is acquired and processed on a computer, and the best architecture with the slightest error was selected to be implemented on the Arduino Nano. The weights and biases for the best architecture were entered in the Arduino Nano and then calculated using the new data from the Electromyograph sensor. The average success rate was 88% for the real-time test.