Abstract: Due to the irregularity of complex intersecting line seams, the welding gun needs to frequently accelerate, decelerate, and change direction during automatic welding, resulting in a certain lag time between receiving instructions, processing data, and executing actions in the welding control system. The welding gun cannot immediately reach the desired position and speed, leading to overshoot and affecting the final welding effect. Therefore, a dynamic compensation method for the motion trajectory of automatic welding of complex intersecting line seams is proposed. Preliminary collection of visual information of the welding gun, using Otus algorithm to extract the main body information of the welding gun end within the video frame, and introducing Mean Shift algorithm to project its two-dimensional position information into vector space. Combined with Kalman filtering algorithm, the motion state information of the welding gun end is estimated to achieve dynamic tracking of the welding motion trajectory. Based on the deviation between the trajectory tracking results and the established welding trajectory, an active disturbance suppression controller is constructed. The Kalman filter outputs the error between the motion trajectory tracking results and the actual complex intersection seam as input. By adjusting the nonlinear factor, the error is nonlinearly compensated to reduce overshoot during the welding process and achieve dynamic compensation of the welding motion trajectory. The experiment shows that the proposed method has accurate trajectory tracking effect and high anti-interference ability, which is of great significance for ensuring the quality of welding operations.