AI-Powered Digital Twin System for Post-Earthquakes Search and Rescue Operations Using Satellite and Drone Imagery

This paper proposes an artificial intelligence (AI) powered digital twin system design to assist the search and rescue process by rapidly analysing high resolution post-disaster satellite and drone images. The system is capable of detecting damaged or collapsed buildings, as well as closed roads. This is achieved through the utilisation of deep learning algorithms, which are trained using high-resolution annotated drone or satellite images. These algorithms constitute a layer that is integrated into a digital twin model. This will enable the digital twin model to optimise the deployment of search and rescue process to damaged areas and determine optimal vehicle routes in real time. A dataset has been developed by integrating open-source annotated datasets, and the model selection and development phase has commenced. The subsequent phases of this work entail the development and training of an AI model in conjunction with the creation and integration of a digital twin, which is to be based on geospatial data services in conjunction with the AI model. Furthermore, a prototype will be tested in simulated earthquake scenarios. The present work delineates the overarching architecture of the system, its objectives, methodologies, and prospective advantages.