In this paper, we propose a systematic approach to identify the friction of an experimental planar two link flexible manipulator. The dynamic model of the flexible manipulator, which contains only the first flexible modes of each link and the model of actuators, has been linearized around a definite operating point. Then, a new concept, "generalized friction", is proposed, and the linearized model is modified to include the generalized friction. A minimal order, linear, time-varying model is obtained. Later, a "strong tracking filter" (STF) is introduced to estimate the time-varying generalized friction at every operating point, based only on a rough model. Although the time-invariant linear model is obtained at a definite operating point, we have shown that it can describe accurately the manipulator's dynamics in the whole operating area with the aids of estimation of the time-varying generalized friction. Finally, experimental results are presented to demonstrate the effectiveness of the proposed approach.

An adaptive augmented state feedback control (AASFC) approach is presented for the two-link flexible manipulator, the Duisburg flexible manipulator 01 (DFM01). In order to describe the DFM01 accurately, two disturbance variables are introduced as augmented states. The controller design is based on the steady-state LQR technique in conjunction with an adaptive compensator, where all states are adaptively estimated by employing the strong tracking filter. This control strategy has been tested on DFM01 for both point-to-point and tracking control of the end-effector with different tip mass. Experimental results show that the controller is quite satisfactory, and has strong robustness against payload variations.