Adaptive Robotic Welding System Using Laser Vision Sensing for Underwater Engineering

Abstract

It's important to equip weld robots with real-time seam tracking technique in order to weld more accurately. In this study, a robotic welding system with a laser sensor, which can detect the position of the V groove, is developed for underwater engineering. The laser sensor includes a laser projector, a camera and laser filter. The laser stripe images are captured and input into a PC real-timely when welding. An improved method of maximization of interclass variance is proposed and is used to segment the laser stripe images. A new approach is developed based on statistical and the syntax approach in welding groove recognition. Welding experiments are implemented on V-grooves. Results show that the system can meet the requirements of general welding seam tracking.