Using GA-based intelligent control means to enhance human-machine interfaces

A GA (genetic algorithm) search procedure was employed to explore a best set of sensory feedback parameters in designing a human-machine interface for improved performance. The optimization concerned two objective functions of interest, which incorporated tradeoffs between speed and accuracy in tracking. A Pareto-optimal front was calculated involving the two cost functions selected This approach differs from the traditional minimum of a non-convex cost function (scalar) describing the desired closed loop performance. Also, this methodology used a parsimonious experimental design method. By making a few runs with a limited number of subjects, a response model was first developed This model was then simulated and a complex vector response surface was generated by the performance variables of interest. The GA search procedure was then used to locate the minimum of this response surface. Finally, in a post hoc experimental study to confirm that the selected design parameters were the best from the class selected, seven human subjects were evaluated at the most favorable experimental design parameters and compared to alternative conditions.