To ensure quality in robotic welding, we need workpieces with a repeatable and consistent shape of the weld seam. In practice, variations occur between workpieces of the same type, to which we adapt with use of seam tracking sensors. As part of the thesis, we developed software for a laser profilometer for the purpose of tracking the weld seam and adapted it to capture and process profiles in real time. We designed testbeds where we tested the software in a laboratory environment and in an environment without welding. We discovered that, with the help of optical and software filters, we can reliably measure the layout of the weld seam using developed software. We identified reflective surfaces, arc light radiation, ambient light, unstable network connection and insufficient power supply as disturbance variables.
|
