Monocular Computer Vision Method for the Experimental Study of Three-Dimensional Rocking Motion

Abstract

The rocking problem is applicable to a wide variety of structural and nonstructural elements. The current applications range from bridge pier and shallow footing design to hospital and data center equipment, even art preservation. Despite the increasing number of theoretical and simulation studies of rocking motion, few experimental studies exist. Of those that have been published, most are focused on a reduced version of the problem introducing modifications to the physical problem with the purpose of eliminating either sliding, uplift, or the three-dimensional (3D) response of the body. However, all of these phenomena may affect the response of an unrestrained rocking body. The intent of this work is to present a computer vision method that allows for the experimental measurement of the rigid body translation and rotation time histories in three dimensions. Experimental results obtained with this method will be presented to demonstrate that it obtains greater than 97% accuracy when compared against National Institute of Standards and Technology traceable displacement sensors. The work concludes with two example experimental studies of rigid body rocking measured with this method as proof of concept. The experimental results highlight important phenomena predicted in some state-of-the-art models for 3D rocking behavior.