Abstract: For the hand-eye calibration problem of the robot platform equipped with a line laser profile sensor, the calibration relying on the 2D point cloud information output by the line laser profile sensor, is of disadvantages of complex calibration process and low calibration accuracy. To solve these problems, a hand-eye calibration method based on cylindrical side-face constraint is proposed. By changing the end pose of the scanning robot, the scanning data of the cylindrical side-face in different poses are obtained. For the elliptical profile obtained by the intersection of the laser plane and the cylinder side-face, the random sampling consensus (RANSAC) algorithm is used to get the coordinates of the center point of the ellipse section. The constrained optimization equation is established by using the distance from the estimated center point of the elliptical profile to the central axis of the cylinder, and thus the hand-eye calibration problem is transformed into the constrained optimization problem. A fusion algorithm of particle swarm optimization (PSO) algorithm and generalized Lagrange multiplier method are used to solve the constrained optimization problem, and the transformation matrix is obtained for hand-eye calibration. Finally, simulation and scanning reconstruction experiments are carried out based on the proposed method. The effects of the error of calibration data, the position of calibration reference object and the initial values of calibration parameters on the calibration results are discussed, and the hand-eye calibration accuracy is verified. The results show that the method is not affected by the position of the calibration reference object and the initial values of the calibration parameters. It is of characteristics of a simple operation, a universal versatility, and a high calibration accuracy. The calibration accuracy is within 0.15 mm, which is suitable for a robot hand-eye in-field calibration.