High-fidelity geometric validation for resilient digital fabrication: A physics-informed multi-modal fusion system demonstrated on complex structural interfaces

The resilience of modern engineering structures, particularly those utilizing modular systems and digital fabrication (e.g., 3D printing), fundamentally depends on the geometric precision of manufacturing and assembly. Deviations in complex structural components, especially at interconnection joints, can compromise load transfer mechanisms and reduce multi-hazard resistance. However, rigorous geometric validation of such complex, reflective surfaces remains a challenge due to sensor alignment ambiguities. To ensure the structural integrity of high-performance components, this paper proposes a physics-informed multi-modal fusion framework for high-fidelity 3D metrology. Using aviation spiral bevel gears—which represent complex structural interfaces with stringent geometric tolerances—as a rigorous test case, we introduce: (1) A robust multi-modal calibration method utilizing Bird’s-Eye-View (BEV) feature encoding to enforce physical geometric consistency; and (2) A two-stage, anti-aliasing deep learning model for extracting precise morphological features. Experiments demonstrate a reconstruction error of less than (Formula presented) and robust segmentation across 35 models. By achieving a measurement accuracy of (Formula presented) with a fivefold efficiency increase, this system provides a critical tool for digital fabrication validation. It ensures that the “as-built” geometry matches the “as-designed” specifications, thereby safeguarding the structural performance and resilience of modern engineering systems.