Auto-calibration of robot workcells via remote laser scanning

Abstract

For new robotic products, the rapid generation of robot programs is generally ensured by offline programming (OLP). Nevertheless, robotic calibration of workcells is essential because of the geometric differences between the three-dimensional (3D) model environment and the real world. In contrast with previous robot self-calibration methods that use specific sensors, the proposed method requires neither marker settings nor initial positioning of the measurement system. Instead, the geometric feature of interplane relations finds relative differences between the coordinates of the 3D model and those of the real workcell. In addition, an iterative method of estimating world coordinates is introduced to increase the accuracy of the revised robot path. A portable laser scanner acquires 3D point-cloud data, so it can be located near the workcell instead of being pre-fixed, thereby facilitating its use. Through experimentation, an average distance of 0.0817 mm is found between the OLP-teaching points and the calibrated points of the pointing operation. © 2022 Elsevier Ltd