Localizing internal steel defects using a piezoelectric sensor system with a robotic hand

초록

Proactive detection of internal defects in steel is crucial for ensuring structural quality and safety. Recent advancements in artificial intelligence and robotics have led to the development in the field of integrated non-destructive testing systems. In this paper, we propose a novel system for detecting and localizing internal defects in steel plates by integrating a miniaturized piezoelectric sensor into a robot hand. Specifically, this system comprises a piezoelectric actuator and sensor on the robot hand. To test our system, steel plates are prepared with internal defects at nine different positions. When the robot hand grasps the steel plate, the actuator generates vibrations, and the sensor measures the frequency response transmitted through the plate. With this sensing data, a machine learning model is trained to find the location of the internal defect. We obtain an accuracy close to 100% with the system. Furthermore, we discuss the feasibility of our system by comparing various learning models and show various applications of this sensing system.

키워드