Friction Identification and Model-based Digital Control of a Direct-Drive Manipulator

Several tasks of the most recent robotics applications require high control performances, which cannot be achieved by the classical joint independent control schemes widely used in the industrial field. The necessity to directly take into account parasitic phenomena affecting motion control, such as friction, often leads to the development of model-based control schemes. The actual effectiveness of such schemes is strongly dependent on the accuracy with which the robot dynamics and the friction effects are compensated by the identified models, and it must be assessed by suitable experimental tests. In this chapter, different solutions are investigated for the development of a model-based control scheme, including joint friction compensation, for a two-links, planar, direct-drive manipulator. In particular, the use of available nominal robot inertial parameters for the identification of a nonlinear friction function, based on the well-known LuGre model, is compared with a complete dynamic calibration of the manipulator, including the estimation of both the robot dynamics and the parameters of a polynomial friction function. The identification results are discussed in the two cases, and inverse dynamics control schemes, based on the identified models, are experimentally applied to the manipulator for the execution of different trajectories, which allow the evaluation of the control performances in different conditions.