AI-Driven Cognitive Digital Twins for Human-Centric Decision Making in Smart Cities

Abstract

Smart cities have become increasingly challenged by the impacts of climate change and rapid urbanization, all of which require decision-making frameworks that support both efficient operation of infrastructure and human health/well-being along with ethically based decision-making processes. Current digital twin technologies that are focused on replicating physical urban infrastructure primarily provide static or data-centric representations of physical assets in smart cities and do not include mechanisms to allow for cognition, adaptability, and/or human-centric decision-support. In this paper, we propose a Cognitive Digital Twin (CDT) for smart cities based upon AI that provides real-time data from IoT sensors, utilizes machine learning to predict/optimize outcomes and includes human-in-the-loop feedback mechanisms to support adaptive, transparent, and context-specific urban decision-making processes. The proposed multi-layered architectural structure for the CDT consists of three layers: a Physical Sensing Layer; a Virtual Modeling Layer; and a Cognitive Intelligence Layer that is integrated with Explainable AI and Policy Constraints. A number of practical applications are provided to illustrate how CDTs can be used to facilitate scenario planning and ethical decision-making when there is uncertainty. Therefore, our main contributions of this paper include: (1) a System Level Architecture for Human-Centric Decision-Making Governance using CDTs in Smart Cities; (2) an Integration Strategy for Collecting Feedback from Humans and AI Models in Digital Twins; and (3) Design Considerations for Aligning Decision Support Provided by CDTs with Trustworthy, Resilient and Citizen-Focused Urban Systems Research Trends identified by IEEE.