Abstract: The calibration of vision sensor and welding robot is essential in the application of intelligent robot. In this article, to solve the hand-eye calibration of a welding robot holding a line structured light sensor, a method based on a spatial straight line is proposed. Firstly, the robot is required to manipulate randomly its end-effector so that the light plane projected by the line structured light sensor could intersect the calibration template at different configurations. Then the characteristic point extraction algorithm is used to extract the intersection point of the line structure light and the straight line. And a nonlinear optimization model is established based on the linear constraint. Finally, the penalty function method and the advanced Powell algorithm are used to solve the position and direction parameters simultaneously. Based on the above method, a series of experiments is carried out and the experimental results show that the method is versatile and suitable for on-site calibration, which effectively improves the calibration accuracy. The calibration accuracy is within 0.2 mm.