Corner Point-based Calibration Technique for QR Code Readers in Robot Localization

초록

In industrial automation, multiple SCARA robots often perform similar tasks within a shared workspace. To reduce setup time and maintain task consistency, it is desirable to reuse teaching data, manually defined Tool Center Point (TCP) positions, from one robot on another. However, mechanical installation differences, coordinate frame misalignments, and sensor placement variations prevent direct sharing of teaching values. This paper proposes a 6D transformation parameter-based calibration algorithm to align the coordinate frames of QR readers mounted on different SCARA robots. Each robot estimates the 6D pose of a rigid zig-shaped QR marker fixed on its base using a robust Perspective-n-Points (PnP) method, with a comparative analysis identifying the Visual Servoing Platform (VISP) implementation as the most accurate and stable. The relative transformation between QR readers is then computed and applied to transform all poses from the target robot's frame into the reference robot's frame. Reprojection error analysis demonstrates that the proposed method significantly improves calibration accuracy, achieving a mean error of 37.35 pixel (7.47 mm) compared to 55.58 pixel without rotation correction and 214.93 pixel for a conventional 2D transformation. This approach enables direct and precise sharing of TCP teaching values between SCARA robots, improving efficiency and consistency in multi-robot industrial environments. © 2025 IEEE.