Advanced solution for a benchmark robot control problem

The paper presents one solution to a benchmark problem, where the target was to control the position of one robot axis. The model given corresponds to a commercial robot, although some simplifications have been made in constructing the model. The nonlinear and time-varying nature of the process together with large disturbances of several types were the challenge for control design. Traditional optimization and frequency domain techniques were used for tuning suitable PID controllers for each specific disturbance type. Based on input-output measurements the disturbance types were identified using a wavelet technique, and the different PID controllers were scheduled according to the identification result. The performance of the developed controller was tested by simulations and by calculating the value of a cost criterion.