Uncertainty Inverse Analysis of Positioning Accuracy for Error Sources Identification of Industrial Robots

Abstract

Uncertainties widely exist in modeling parameters of industrial robots due to wear during service. These uncertainties are too small to be directly measured and may severely affect the positioning accuracy of end-effector through multi-stage propagation. An uncertainty inverse analysis method of positioning accuracy is proposed to identify uncertainty of typical structural parameters with the selected optimal position information. The Denavit–Hartenberg method is employed to establish the kinematic model. For the probability uncertainty of modeling parameters, the inverse analysis model, which reflects propagation mapping between uncertain variables and end-effector position response is obtained by first-order second-moment method. The partial-derivative information is used to improve calculation efficiency through removing low-contributing parameters. To further reduce measurement noise interference, orthogonal matching pursuit strategy is presented to search optimal sensor placement with the objectives of maximum coefficient modulus and minimum correlation, then pseudoinverse matrix method is employed to obtain the exact solution of identification equation. Finally, numerical simulation results for 6 degrees-of-freedom robots indicate that the proposed method with improved solution efficiency can accurately identify uncertainties of joint angles within the error 5% under large measurement errors.