Development of Computer Vision Method for Monitoring a Dynamic Pile Loading Test

Abstract

This study develops a computer vision method for monitoring pile behaviors during pile load tests compared to dynamic pile load test and Case Pile Wave Analysis Program (CAPWAP). The detection was carried out by the use of ArUco markers attached to the rammer and the head of the pile. The markers can be used to measure the distance in 3-D between the camera lens and itself. The ArUco recorded videos were analyzed by developed code with OpenCV library in Python. The Markers movement at rammer and pile head will be analyzed to obtain force and velocity. The pile capacity was estimated by the Case method. The accuracy of the measured displacement depends on the distance between the camera and the markers. The smaller the pixel of the markers, the greater the error. The calculated velocity is similar to DLT results. However, there was significant error for calculated forces due to cumulative marker movement and velocity detection error, including misjudging wave speed and pile modulus. This study presented the possibility of using computer vision to monitor DLT.