RoboTuni: An Intelligent Servo-Tuning for Improving Path Accuracy in Robot Manipulators

This article proposes an intelligent diagnostic and servo-tuning approach for servo-mismatch issues of six-axis robotic manipulators to improve the dynamic path accuracy, referred to as the RoboTuni. At first, a virtual servo-drive system is developed to simulate and assess the servo-drive system’s performance. To diagnose servo mismatches, a 1-D convolutional neural network (1D-CNN) is adopted to analyze trajectory tracking errors and identify the corresponding poor performance servo axis. Subsequently, a Lagrange interpolation-based tuning method is proposed, enabling efficient parameter adjustments without requiring large datasets or extended convergence periods. Simulation and experimental results are introduced to demonstrate that the proposed approach significantly enhances the dynamic responses of the joint servo and improves the overall path precision of the system, including a 67.38% increase in mean path accuracy in circular motion and a 69.41% maximum path error reduction in linear motion of the Y-axis.