摘要
Due to its high production flexibility, roller hemming has become the mainstream process for forming and joining metal sheets in the automotive industry. The traditional roller hemming process requires specific dies to support sheet metal parts and repeated offline manual adjustment of hemming routes, resulting in high die costs, high time consumption, and excessive labor inputs. The universal platform presented in this paper could replace specific dies to effectively reduce costs and expand production flexibility. To reach this objective, a vision-based automatic compensation path to achieve a dies-free roller hemming process is proposed and investigated in this paper. Hand–eye sensor modules assisted by multi-coordinate synchronization calibration for the roller hemming were designed to reconstruct three-dimensional (3-D) shape data of the incoming materials. Results from the proposed system were validated with experimental measurements for the sheet offset and the compensation of the arm hemming position, showing that the single-axis error can be reduced to ≤0.1 mm.