An edge computer based driver monitoring system for assisting safety driving

Abstract

Driver Monitoring System (DMS) is a promising IoT application in Intelligent Transport Systems (ITS) research field. DMS assists car drivers by monitoring their driving activities, sensing incidents to cause possible dangers, and alerting the drivers to prevent accidents. We aim to realize a new DMS that is inexpensive and highly effective. This paper proposes a method for detecting any incidents based on machine learning. The proposed method firstly configures a detector by training in-car environment data and driver’s vital signs gathered from multiple sensors. Then, the detector is embedded in a self-contained edge computer for monitoring a driver in a car. The device is always connected to the information communication network by radio waves. Those data obtained by monitoring are stored in the cloud server. The server learns and analyzes the stored data using processing such as machine learning. As a result, we acquire knowledge leading to safe driving. The edge computer uses these knowledge to process the sensor data in real time, observe the driver, sense the danger, and call attention. These mechanisms prevent occurrence of troubles such as traffic accidents. The paper describes the proposed system overview, implementation method, and initial evaluations.