Computer-aided design of nonlinear controllers for torque controlled robot arms

Abstract

The paper extends a design methodology for designing nonlinear controllers for torque-controlled robot arms. The controller consists of a nonlinear precompensator which makes the robot arm appear as a set of linear second-order systems (one linear system for each degree of freedom) and a linear feedback controller. The nonlinear controller is parametrised by the desired closed-loop transfer function, which is linear. This automatically achieves stability of the nominal system. The parameters of the controller, which are parameters, such as bandwidth, of the desired closed-loop transfer function, are then adjusted to satisfy engineering constraints such as hard constraints on the control (which is a nonlinear function of the state), robust performance and robust stability (maintenance of performance and stability despite variations in the parameters, such as friction and mass, of the robot arm). It is shown how recently developed optimisation-based design methods may be employed. An example is considered, and simulation results presented to illustrate the design procedure.