Teleoperated welding by learning welder's torch maneuvering based on the wrist-inspired parallel mechanism

Tungsten Inert Gas (TIG) welding is a critical process for field repair of spent fuel pools in nuclear power plants. The fit-up accuracy of welded joints is crucial for preventing weld defects for robotic TIG welding of the root pass. However, it is challenging to obtain welded joints with consistent fit-up due to large structural deformation. The root pass welding with oversized gaps poses a significant challenge for field robotic welding. To address this challenge, this paper proposes a teleoperated welding method based on the wrist-inspired parallel mechanism (WPM) by learning welders' torch maneuvering. A novel wrist-inspired parallel mechanism is developed based on learning expert welders' torch control behavior. It achieves higher torch control speed than welders. Moreover, the teleoperated root welding method is proposed based on the WPM. The experiments are conducted to evaluate the teleoperation and root welding performance of the proposed system. The root welding experiments are conducted under the challenging condition of an oversized gap. The results demonstrate that the proposed parallel mechanism achieves superior control speed of the welding torch compared to manual welding operations. The pose control speed reaches 272°/s, exceeding that of welders by 27 %. It overcomes a key limitation in the rotational speed of traditional industrial robots. Moreover, the proposed teleoperated welding method achieved high quality root welding with an oversized gap of 5 mm. It overcomes the challenge of oversized joint gap, enabling robotic root welding in the challenging field repair conditions.