<p>Achieving precisely accurate joint preparation geometry is critical to enable welding conditions for high-quality robotic welding. Recently, smart multi-robot co-operation has enabled more flexibility and accuracy in positioning the components of a weldment for fit-up, tack welding and welding. Therefore, smart multi-robot welding could improve process cycles and has potential to revolutionize low-volume high-variation production. Currently, welding distortions cause angular and axial misalignments in plate components, thus, creating a significant challenge for smart multi-robot welding production of welded structures. However, the impact of angular welding distortion could be mitigated by pre-setting components during material handling preparations. Still, the pre-set angle is typically unknown, which complicates the path and process planning of multi-robot welding during the setup task. The objective of this research is to develop a smart multi-robot welding procedure that advantageously exploit angular welding distortion to achieve a high-quality welded joint. The applied methods in this study to determine pre-set angle for joint preparation consist of numerical and empirical welding experiments on T-joint, corner joint and single-V butt joint test-specimens. The angular distortion is measured with an optical sensor from the test-specimen. The study aims in contributing to a development of a novel feedback system, where the sensor data is utilized to calculate the correct manipulator position in the digital twin of the multi-robot station to adjust the optimal pre-set angle. Then verification experiments are conducted by applying pre-set angle to test-specimens. Finally, the results are compared to the angular misalignment tolerance limits of the EN ISO 5817 and EN ISO 13920 standards to verify the performance of the developed feedback system. The study indicates that the smart multi-robot welding procedure performs well on the studied joint types.</p>