Recognition of human arm gestures using Myo armband for the game of hand cricket

Abstract

Gesture Recognition is the most recent development in the field of Bio Robotics. The proposed paper focuses on presenting a low cost sensor based human gesture recognition for the game of Hand Cricket. Hand cricket is a popular game in south Asian countries which involves the use of human finger gestures to score. This game is generally played between two players. Each player has a pre-defined gestures for the scores one, two, three, four and six. Both the players are made to wear the Myo armband. Myo armband is used to capture the Bio-potentials triggered during every muscle action. The various gestures performed in this game triggers various muscle group signals. A data set is created by collecting the eight channel bio potentials for every gesture made by both the players. The obtained data set is pre-processed and feature extracted. Now the Machine Learning techniques are performed in the data set to classify all the five different gestures with maximum accuracy. Support Vector Machine (SVM) gave the maximum accuracy to the classify the data set of both the players. The efficiency obtained for both the players are 92% and 84%. The proposed system is made to train with the data set obtained by the two players and the game is played in real time with the help of two MATLAB in two computers. Along with the classification of data, the scores of the individual player is calculated and displayed. With the scores being displayed, we can determine the player who scored the highest and the winner of the game can be determined.