Online identification and robust compensation of extended nonlinear time-varying friction model in robotic arms

Torque loss values resulting from friction in mechanical arm joints and their actuators were identified. Then, using the LuGre friction model and applying the nonlinear least square error estimation method, a more precise model is presented to model its effect. To investigate more efficiently, the general LuGre friction model was used in three forms for modeling the system’s friction. The first model was the common LuGre model, while in the two other extended models, the parameters were considered time-varying that are estimated online. Of note is that the third model also includes the nonlinear viscous term. Then, by utilizing a nonlinear sliding mode controller, simulations were performed for both point-to-point and linear path of the system’s state, and the efficiency of designed algorithm was examined. The results show the advantage of the identified models with time-varying parameters over the constant parameters model. Eventually, a 6R laboratory robot was employed for validation of the proposed algorithm, subsequently validating the simulation results, and as it showed, the final friction model presented better performance.