Host: The Japan Society of Mechanical Engineers
Name : [in Japanese]
Date : June 05, 2019 - June 08, 2019
A program was developed to recognize the movement of a hand using IMU (Inertial Measurement Unit) sensor, which is also used by smartphones. A sensor module was made by connecting the IMU sensor with a pressure sensor. Then, in order to obtain an accurate recognition rate with CNN (Convolutional Neural Network), a signal image was made from the gesture signal data collected using the IMU sensor attached to the hand. The neural network model was trained with multiple gestures including handwaving, beckon, circular motions. The optimization process resulted a high test-accuracy of 97%. Furthermore, the program was also tested to recognize gestures in real-time by utilizing the pressure sensor as a switch. The experimental results confirmed that hand gestures can be recognized with excellent accuracy.
