Abstract
The Experience-Enhanced Material-Efficient Generative Bridge Design (EMBED) project seeks to innovate the bridge design process by leveraging generative AI to produce material-efficient and structurally sound solutions. Addressing the inherent complexity and variability of bridge engineering, EMBED combines data-driven methods, engineering knowledge, and practical expertise to automate and optimize the design process. The project will develop an interactive, multimodal AI framework that utilizes deep learning, reinforcement learning, and physics-informed loss functions to embed human experience and engineering constraints directly into the generative models. Key activities include the collection and annotation of diverse bridge data, the creation of a comprehensive knowledge graph, and the development of a functional prototype for AI-assisted bridge design. The workflow is designed to support iterative human-AI collaboration, allowing experts to guide and validate the outputs of generative models. A student bridge design competition will further demonstrate the practical impact of the project, providing real-world validation and engaging the next generation of engineers. Through these innovations, EMBED aims to set new standards for efficiency and creativity in bridge design.
