Using manual tracking performance to tune a specified-negative-amplitude input shaper

This paper proposes a method for optimizing a control system that includes a human in the loop. The method uses human-operator testing to measure the system´s performance for a given value of a control parameter. Root-mean-squared (RMS) tracking error is used to quantify performance. The optimum control parameter value is found using a one-dimensional search algorithm. As an example application, the golden-section search method is used to find the optimal form of a specified-negative-amplitude (SNA) input shaper. Results from an operator experiment show that the method was able to identify an optimal range of parameter values using a relatively small number of iterations.