Collaborative Cloud-V. Edge System for Predicting Traffic Accident Risk Using Machine Learning Based IOV

Smart city development is profoundly impacted by cutting-edge technologies such as information and communications technology (ICT), artificial intelligence (AI), and the Internet of Things (IoT). The intelligent transportation system (ITS) is one of the main requirements of a smart city. The application of machine learning (ML) technology in the development of driver assistance systems, has improved the safety and the comfort of the experience of traveling by road. In this work, we propose an intelligent driving system for road accident risks prediction that can extract maximum required information to alert the driver in order to avoid risky situations that may cause traffic accidents. The current acceptable Internet-of-vehicle (IOV) solutions rely heavily on the cloud, as it has virtually unlimited storage and processing power. However, the Internet disconnection problem and response time are constraining its use. In this case, the concept of vehicular edge computing (V.Edge.C) can overcome these limitations by leveraging the processing and storage capabilities of simple resources located closer to the end user, such as vehicles or roadside infrastructure. We propose an Intelligent and Collaborative Cloud-V.Edge Driver Assistance System (ICEDAS) framework based on machine learning to predict the risks of traffic accidents. The proposed framework consists of two models, CLOUD_DRL and V.Edge_DL, Each one complements the other. Together, these models work to enhance the effectiveness and accuracy of crash prediction and prevention. The obtained results show that our system is efficient and it can help to reduce road accidents and save thousands of citizens’ lives.