Track: High School STEM Poster Competition

Abstract

In this paper, we present an EMG signal analysis and classification technique for hand gesture recognition specialized for autonomous vehicle environments. In addition, we propose a new user interface method that considers the bio signals received from wearable devices and the characteristics of self-driving cars. EMG is an electrical property produced to command the contraction and relaxation of muscle tissue. The medical surface EMG sensor mounted on the MYO armband used in the experiment transmits EMG signals and sensor information wirelessly from the arm to the terminal via Bluetooth. The EMG-based interaction system proposed in this paper processes the collected EMG signals to classify hand gestures and interacts with the autonomous driving system. Gestures for control transfer are classified in a double threshold method through EMG signal analysis and used as a more efficient interface by applying a separate low pass filter. This EMG-vehicle interaction system was developed to enable control of an autonomous vehicle through a system capable of controlling an autonomous vehicle. In addition, we introduce constraints and methods using vehicle state information to reduce gesture recognition errors that may occur in situations unintended by the driver.

Keywords

EMG sensor, wearable devices, EMG-vehicle interaction, autonomous vehicle and Data Augmentation.