Digital Twin: Challenge Road Damage Detection on Edge Device

Abstract

The development of digital twins as a new technology aligns with the main strategic objectives of some of the world's top manufacturing countries. Several studies on digital twin cities have been performed, including the interaction of the Internet of Things (IoT) with smart cities, smart city traffic disaster risk management, and smart road inspection. The purpose of the research is to offer a survey of digital twin cities that includes road damage inspection as support for smart roads. The deployment of IoT technology and artificial intelligence (AI) to construct road defect detection systems on edge devices. This is the most recent breakthrough and an open challenge to creating a road defect detection system using edge AI on digital twins. As a result, edge AI devices are required to perform automatic and real-time road inspections. However, edge AI devices for detecting road damage have limited computing and storage capacities. Much of the physical system modeling has been completed in the construction of digital twin cities for road damage detection. In order to deploy CNN models on devices, some require physical system development. The stages involved in the development and implementation of a CNN model for an edge AI device road damage detection system are hardware selection, software development, developing the road damage detection application, edge device optimization, security and connectivity, testing and validation, and deployment. To improve system performance and reduce inference time, the CNN model must be optimized before being deployed on edge devices as edge AI.