Abstract： In order to achieve robot tracking of vertical welding seam accurately, improve the quality and efficiency of welding, and reduce labor intensity of welders, the sampling current was filtered by method of combination. Linear regression was done by using the method of “absolute difference” from the first sampling point to the 32th sampling point to regard the slope of fitting line as a deviation. The deviation and deviation change rate was used as the input of the fuzzy controller. The vertical welding seam was tracked by controlling level slider telescopic after fuzzification of the input, fuzzy reasoning and defuzzification of the output. The welding torch was made to move from updip 45 degrees to downdip 45 degrees by path planning. The vertical welding seam was tracked by combining path planning and horizontal slider telescopic. Experimental results show that the robot can well track the vertical welding seam based on rotating arc sensor by using the algorithm, which have good reliability and high accuracy.