In traditional air-conditioning plants, leak detection task is usually done manually. Manual leak detection requires the use of measuring instruments, which are complicated to operate, and workers often fail to implement standard operating procedures (SOP), resulting in negligent detection and high missing rates of leak detection. Manual leak detection is also less efficient due to the unobvious characteristics of welding spots.
So the client expects robots to recognize several welding spots in different positions on the copper tube and direct the muzzle of leak detector to the spot position for refrigerant leak detection. It is difficult for traditional robotic solutions to help clients solve problems due to the complex pipeline of the corresponding area, the random location of welding spots, the lack of visible welding spot characteristics, and the wide variety of parts.
When adopting teaching method rather than vision, there will be a 2-3 cm deviation from the top, bottom, left and right each time the welding spot is displayed, and the robot cannot carry it out with a fixed route.