A new iterative identification algorithm for estimating the LuGre friction model parameters




Mahmoudkhani, Saeed
Gorenstein, Johnathan
Ahmadi, Keivan

Journal Title

Journal ISSN

Volume Title


Mechanism and Machine Theory


The parameters of dynamic friction models like the LuGre model are commonly identified by computationally intensive nonlinear optimization methods. In this paper, an alternative least-square-based iterative algorithm is proposed to simultaneously identify the LuGre model parameters and the inertial properties from a limited set of measurements. The LuGre model, expressed in two forms allowed independent identification of the static and dynamic parameters. Moreover, since the method uses response-input time-history instead of numerous constant velocity experiments (CVEs), both the inertial and friction parameters can be identified in much fewer experiments (theoretically one). A variant of the Sparse Identification of Nonlinear Dynamics method called SR3 is embedded in the algorithm to capture the nonlinear viscous friction and Stribeck effects in the friction model. Application of the algorithm to an industrial robot joint shows that the convergence of the algorithm is fast and the identified model is accurate in predicting the joint’s dynamics in a wide range of velocities. The friction-velocity curves resulting from the identified model are compared to those obtained by traditional CVEs to confirm the accuracy of the identified model.



LuGre friction model, Sparse regression, SINDy-SR3


Mahmoudkhani, S., Gorenstein, J., & Ahmadi, K. (2023). A new iterative identification algorithm for estimating the LuGre friction model parameters. Mechanism and Machine Theory. Preprint.