Enabling cognitive and autonomous infrastructure in extreme events through computer vision

Abstract

As advent by the continuous inertia toward integrating artificial intelligence into daily operations, it is a matter of time before artificial intelligence reforms the field of structural engineering. From this point of view, this paper explores how computer vision and deep learning can be applied, in combination with advanced finite element analysis, to realize cognitive (self-diagnosing) and autonomous infrastructure. The outcome of this study demonstrates that computer vision not only can enable a structure to understand that it is undergoing an extreme event but can also allow it to trace its own performance and to independently respond to mitigate prominent failure/collapse. Findings of this work infer that computer vision can serve as an intelligent, and scalable agent to accurately trace structural response, identify different damage mechanisms and propose suitable repair strategies whether during or in the aftermath of a traumatic event (i.e., fire, earthquake). Finally, a series of challenges and future research directions are outlined toward the end of this paper.