An Efficient Parameter Identification Framework: A Case Study on Robot Manipulators

Recently, robot applications in manufacturing have been becoming more challenging. Accuracy of the dynamic parameters of the robot is demanded for the advanced robot controller design. This article proposes a novel parameter identification method for the simultaneous estimation of linear and nonlinear parameters. The parameter identification problem of a robot manipulator is considered as the study case. The dynamic parameters are divided into linear and nonlinear sets. The inverse dynamics model (IDM) is reduced into a form linear in relation to the linear set, while the nonlinear set remains in the regression matrix. A global optimization problem is formulated for the parameter identification. The multistart (MS) algorithm is incorporated with the least-squares (LS) method to solve the nonlinear identification problem. By innovatively unifying nonlinear and linear identification techniques, the burden of nonlinear optimization is minimized through the integration of a linear regression algorithm, significantly reducing the search space. The proposed framework offers improved accuracy and efficiency in parameter estimation without requiring additional experiments. Furthermore, the global optimization technique improves global searchability, reducing the likelihood of the solution becoming trapped in local optima. The effectiveness of the proposed identification framework is highlighted through experiments carried out by a five-degree-of-freedom (DOF) robotic experimental platform. Compared to two existing state-of-the-art methods, the proposed method achieves up to a 9% improvement in accuracy with two times faster than computation time, underlining the effectiveness and efficiency of the proposed parameter identification framework.