55 A Vision-Based Kinematic Calibration for Pick-and-Place Robot

Pick-and-place robot is widely used in sorts of machines especially semiconductor equipment. Placement accuracy is one of the most important performance criteria of it. In order to improve the placement accuracy cost-effectively, a new approach for kinematic calibration is developed utilizing a machine vision system. Using homogeneous coordinate transformation, each axis of this robot is modeled based on rigid body kinematics and small angle approximation of the errors. A simplified expression of kinematic calibration equation for this robot is deduced based on the character of error components. The coefficients of equation are derived using the least square method (LSM) based on experiment results obtained by the camera acquiring incremental positions of a fixed cross mark. The placement accuracy is improved sharply from ±50um to better than ±30um after kinematic calibration. Result shows this approach is effective and very suitable for pick-and-place robot with machine vision system inside.