Open Access

Roman Haas, Thomas Wendt

• High-precision robotic machining and assembly require accurate calibration of the tool center point (TCP), as small deviations directly propagate into process errors. We present a novel vision-based TCP calibration method using a single low-cost 2D camera, eliminating the need for markers or manual interaction and enabling fully automated calibration. A mathematical approach based onHigh-precision robotic machining and assembly require accurate calibration of the tool center point (TCP), as small deviations directly propagate into process errors. We present a novel vision-based TCP calibration method using a single low-cost 2D camera, eliminating the need for markers or manual interaction and enabling fully automated calibration. A mathematical approach based on transformation matrices was derived to model the calibration process. Experiments were conducted on a real robot, including comparative measurements to validate the accuracy of the results. The proposed method was also compared with a well-established manual TCP calibration, demonstrating a remaining position error of 0.240 mm and an angular error of 0.183°, significantly improving the accuracy. This highlights the potential of the method to enhance reliability and reduce calibration effort in industrial robot applications.…