Abstract
Adaptive structures are an emerging class of load-bearing systems characterized by the integrated co-design of structural layout and control mechanisms, enabling purposeful modification of mechanical response through sensing, computation, and actuation. As such, they fall within the broader category of cyber-physical systems, in which physical and digital components interact through feedback. Within this paradigm, structures sense external actions and actively modulate stiffness and geometry to control deformations and redistribute internal forces. Structural behavior is therefore governed not only by material properties and geometry, but also by embedded computational decision-
making.

​

Recent advances have extended adaptivity beyond real-time response toward structural synthesis and lifecycle resilience. When structural topology and control system layout are co-designed through All-In-One optimization formulations, adaptivity enables new limits of material efficiency that conventional structural systems cannot achieve [1]. At the building scale, adaptive floor systems demonstrate material savings exceeding 60% compared to conventional flat slabs [2]. At the infrastructure scale, design and retrofit strategies for bridges using active components show potential to significantly reduce damage accumulation and extend service life [3], [4]. These principles have been validated through large-scale demonstrators, including a 10 × 6 m rib adaptive floor slab developed at the University of Stuttgart.

​

This keynote reflects on the trajectory from conceptual frameworks to full-scale validation and discusses how the next phase will merge the mechanics of adaptive structures with artificial intelligence for design and operation. Machine-learning models, including Graph Neural Networks, Variational Autoencoders, and Reinforcement-Learning Agents, open new directions for structural topology optimization and control synthesis, enable cross-scale generalization, and embed physical constraints directly into learning architectures. This convergence is expected to uncover previously unexplored, highly efficient structural configurations and to advance adaptive systems capable of self-diagnosis, damage mitigation, and performance optimization. A forward objective is the transition toward “metastructures”, in which adaptivity is embedded across structural, component, and material scales through architected material systems. We are entering a stage where structures will not only respond to loads but will infer their effects and adapt accordingly. The fusion of mechanics and data-driven will play a defining role in the next era of structural design.