Empowering Cognitive Digital Twins with Generative Foundation Models: Developing a Low-Carbon Integrated Freight Transportation System

Effective monitoring and operation of freight transportation ssytem is crucial for advancing sustainable, low-carbon economies. Traditional approaches, which rely on discrete simulations of each mode using single-modal data and simulation, are inadequate for optimizing intermodal and synchromodal systems in a holistic manner. These systems involve complex, interconnected processes that impact shipping time, costs, emissions, and socio-economic factors. Developing digital twins to enable real-time situational awareness, predictive analytics, and optimization of urban logistics systems often demands extensive efforts in knowledge discovery, as well as significant time for integrating various datasets, coupling multi-domain simulations, and developing key software components. Recent advancements in generative AI present new opportunities to create foundational models for scientific research, significantly streamlining the development of digital twins. These models extend traditional digital twins' capabilities by automating the processes of knowledge discovery, data integration, and management to generate innovative simulation and optimization solutions. They also promote autonomous workflows for data engineering, analytics, and software development. This vision paper proposes an innovative paradigm that leverages the power of generative foundation models to enhance digital twins for urban research and operations. Using the decarbonization of the integrated freight transportation as a case study, we propose a conceptual framework that employs recent generative AI advancements, specifically transformer-based language models, to enhance an existing urban digital twin through the development of foundation models. We present preliminary results and share our vision for a more intelligent, autonomous, general-purpose digital twins for optimizing integrate freight transportation system ranged from from multimodal to synchromodal paradigms.