Session: 15-01: Robotic and Automated NDE

Paper Number: 147656

147656 - A Multi-Scale Robotic Approach for Nondestructive Evaluation of Surface Cracks on Concrete Structures 

Abstract: 

Structural defects, such as surface cracks, have considerably smaller scales compared to the large-scale size of most civil infrastructures. This presentation introduces a novel multi-scale robotic approach to address this scale gap. The idea is to imitate the approach of professional human inspectors who first use visual inspection to identify regions of interest and then use more advanced and application-specific nondestructive evaluation (NDE) to map even the smallest defects with higher accuracy and resolution. Specifically, our approach first uses a U-Net neural network to initially identify regions of interest (ROI) containing potential surface cracks. Then, a robotic arm uses a high-definition laser scanner to autonomously scan the geometry of these ROIs. The high-resolution scans will next be fused with a coarse 3D point cloud obtained from a LiDAR scan of the surrounding environment. The approach was validated through both computer simulations in the robotic operation system (ROS) and experiments carried out on a physical concrete slab. For the experimental validation, laser measurements were compared with the readings of transparent crack rulers. The results show an unprecedented resolution of 0.004 mm for crack width measurements for a crack with a width of 0.17 mm. Such results demonstrate the potential impact of the proposed method for assessing the condition of civil infrastructures and quantifying their defects.

Presenting Author: Arvin Ebrahimkhanlou Drexel University

Presenting Author Biography: Dr. Arvin Ebrahimkhanlou is an Assistant Professor in the Civil, Architectural and Environmental Engineering Department at Drexel University with a courtesy appointment in the Mechanical Engineering and Mechanics Department. He received his Ph.D. from and completed his post-doc at The University of Texas at Austin. Dr. Arvin’s research is in the area of robotic-based and Artificial Intelligence (AI)-based assessment of civil infrastructures as well as mechanical and aerospace structures.

Authors:

Ali Ghadimzadeh Alamdari Drexel University
Arvin Ebrahimkhanlou Drexel University