Abstract: To solve the problems of low efficiency, high contamination risk, and poor adaptability in the crucible cleaning and transfer process of the robotic unmanned testing system in the coal quality analysis industry, which are caused by reliance on manual work or semi-automatic equipment, an integrated crucible cleaning and transfer device based on visual guidance is designed. The device is constructed with three core modules: visual perception module, mechanical execution module, and cooperative control module. The visual perception module adopts a combination of an industrial camera and a ring light source, and realizes crucible image recognition, feature extraction, and judgment of stain and damage conditions through an improved YOLOv8 algorithm. The mechanical execution module is composed of a mobile module and a translational cylinder, equipped with a vacuum suction device and a crucible rack to complete crucible movement, operation, and cleaning of different types of crucibles. The cooperative control module connects with the system’s upper computer scheduling through software to realize crucible receiving and dispatching, local cleaning, and status analysis. Experimental results show that the device has a cleaning success rate of 99.2% for different types of crucibles. The mass of residual pollutants after cleaning is ≤0.05 mg, and it can operate continuously for 1000 times without failure. Its efficiency is about 4 times higher than that of traditional manual cleaning.In conclusion, the device can effectively reduce the risk of cross-contamination and improve the overall efficiency of the robotic unmanned testing system. It can be integrated into various robotic unmanned testing systems, providing key equipment support for the automation and intelligent upgrading of testing processes.