Evaluation of Friction Models for Haptic Devices

In this work different friction models are evaluated to determine how well these models are suited for performance simulation and control of a 6-DOF haptic device. The studied models include, Dahl model, LuGre model, Generalized Maxwell slip model (GMS), smooth Generalized Maxwell slip model (S-GMS) and Differential Algebraic Multistate (DAM) friction model. These models are evaluated both numerically and experimentally with an existing 6-DOF haptic device that is based on a Stewart platform. In order to evaluate how well these models compensate friction, a model-based feedback friction compensation strategy along with a PID controller were used for position tracking accuracy. The accuracies of the friction compensation models are examined separately for both low-velocity and high-velocity motions of the system. To evaluate these models, we use criteria based on fidelity to predict realistic friction phenomena, easiness to implement, computational efficiency and easiness to estimate the model parameters. Experimental results show that friction compensated with GMS, S-GMS and DAM models give better accuracy in terms of standard deviation, Root Mean Squared Error, and maximum error between a reference and measured trajectory. Based on the criteria of fidelity, ease of implementation and ease to estimate model parameters, the S-GMS model, which represents a smooth transition between sliding and pre-sliding regime through an analytical set of differential equations, is suggested.