A vision-guided collaborative laser scanning method for additive manufacturing inspection

Abstract Laser ultrasonic testing is widely employed for defect detection in additive manufacturing (AM) components. Conventional scanning methods, however, are often limited by predefined paths when inspecting large-scale workpieces, preventing single-pass coverage. To address this challenge, we propose a vision-guided collaborative laser ultrasonic inspection system. The system integrates both excitation and detection probes onto a si x -axis robotic arm, enabling flexible and adaptive scanning trajectories. For partitioned scanning of large components, a point cloud clipping strategy based on surface geometric features is developed, which constructs a feature-based coordinate system from the point cloud boundary and divides the cloud into equal segments within this system. Additionally, a multi-tool end-effector and supporting middleware are implemented to establish transformation relationships among the feature, camera, robotic tool, and robot base coordinate systems, enabling automated and adaptive laser scanning. Experimental results demonstrate that the proposed point cloud clipping approach accurately establishes the feature coordinate system compared with conventional methods. The system achieves positioning accuracy of approximately 1 mm in all directions. Pointwise excitation and multipoint reception experiments further validate the system’s capability to reliably acquire ultrasonic signals. Overall, the proposed method simplifies operation and overcomes limitations associated with current laser scanning techniques, offering a practical solution for high-precision inspection of AM components.