To address the issues of insufficient degrees of freedom, low intelligence, and poor adaptability to complex welding conditions in existing all-position automatic pipeline welding equipment, a mobile collaborative robotic arm pipeline all-position welding robot system based on laser vision sensing was proposed.The system integrated a mobile chassis, a six-axis collaborative robotic arm, a laser vision sensor, and a flexible guide rail, forming a seven-degree-of-freedom welding system with multi-posture welding torch adjustment capability.A multi-task heterogeneous control system architecture based on a soft PLC real-time kernel was constructed to achieve centralized control of EtherCAT/CANopen multi-bus networks.Seam searching and real-time tracking were performed using the laser vision sensor, and an improved RANSAC(random sample consensus) algorithm combined with an adaptive threshold strategy was employed to achieve robust seam trajectory fitting and dynamic compensation.Experimental results of narrow-groove pipeline all-position welding show that the system design is reasonable, the robot operates stably, the seam tracking accuracy meets requirements, and the weld formation is satisfactory, fulfilling the demands of all-position welding conditions.