Physically Guided Multiscale Visual Servoing Robot for Automatic Detection of Large Heterogeneous Composite Structure

• Published in: IEEE transactions on instrumentation and measurement Vol. 74; pp. 1 - 11
• Main Authors: Zeng, Yu, Kang, Yukuan, Gao, Bin, Liu, Lei, Zhao, Xiaoming, Ren, Hongjiang, Li, Jiacheng, Lok Woo, Wai
• Format: Journal Article
• Language: English
• Published: New York IEEE 2025; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Accuracy; Automatic defect detection; Cameras; Composite materials; Composite structures; Defect detection; Defects; large-scale heterogeneous composite materials; Learning; Manipulators; Mobile robots; physically guided; Platforms; Robots; Solid modeling; Thermography; Three-dimensional displays; Visual control; visual servo; Visual servoing
• ISSN: 0018-9456; 1557-9662
• DOI: 10.1109/TIM.2025.3555676

Optical pulsed thermography (OPT) is an effective method for detecting defects in composite materials and has been widely applied in aerospace and other industries; however, when inspecting large-scale heterogeneous composite materials, conventional fixed detection platforms (which are stationary and unable to change their ground position) cannot automatically and accurately detect defects within a 3-D model in a single operation. This article proposes a multiscale visual servo detection framework based on a mobile detection platform, transitioning from a fixed detection platform with a mobile composite material to a fixed composite material with a mobile detection platform. The defect detection process for large-scale heterogeneous composites is divided into three scales: 1) rapid positioning of composite materials using process learning; 2) precise positioning to minimize system errors and enhance 3-D model accuracy through self-learning; and 3) defect detection via infrared measurement field division. The proposed framework enables fully automatic defect detection, precise defect mapping, and accurate 3-D modeling of large heterogeneous composites. Compared to traditional fixed detection platforms, this approach significantly improves efficiency by detecting large-scale heterogeneous composites in a single operation, achieving high-performance defect detection and enhanced 3-D model accuracy.