Towards Edge Intelligence : Real-Time Driver Safety in Smart Transportation System

Smart Transportation System Md. Shirajum Munir, Sarder Fakhrul Abedin, Kitae Kim, and Choong Seon Hong Department of Computer Science and Engineering, Kyung Hee University, South Korea. { munir, saab0015, glideslope, cshong }@khu.ac.kr Abstract In this research, we introduce an edge intelligence model for real-time driver safety in the smart transportation system to enhance the fifth-generation (5G) networks to the beyond 5G. In order to do this, first, we design an intelligent model for road side unit (RSU) that facilitates driver activity recognition in the edge of the networks, where we adopt the concept of capsule network. Second, utilizing this model, we propose a real-time safety notification algorithm for RSU, where this algorithm is capable of sending real-time safety notification to the vehicle driver as well as to centrally controlled road safety agent. So, the risk of road accident due to distracted driving is minimized. Third, we implement our own environment to validate the proposed model, where the benchmark state farm distracted driver detection dataset is used for the model training. Finally, we show that the higher accuracy in classification of the driver activity justifies the performance of the proposed method with a less computational complexity safety notification algorithm.