Artificial neural network based geometric error correction model for enhancing positioning accuracy of a robotic sewing manipulator

Abstract In order to meet the requirements of extensive and fast fashion changes in the customer demand, there is a need for high flexibility in automation of sewing process. Currently, industrial robots are involved to perform skilful sewing tasks in garment manufacturing; however, it is difficult to achieve the required precision during sewing of typical of geometric shapes in the fabrics due complex mechanical behaviour of fabric materials and geometric errors in the robotic links. In order to control sewing path and its deviation from the desired trajectory, a neural network based approach for predicting the position error correction of a 2R robot manipulator using inverse and forward kinematics models. A simulation study has been performed to investigate the effect of geometric errors of the link lengths on the positioning accuracy of the manipulator in MATLAB environment. Performance of the proposed approach for improved tracking of typical geometric shapes such as circle is demonstrated using simulation results.