Manual Control of Vehicles with Time-Varying Dynamics

An experiment was conducted to seek the models of the human pilot controlling a vehicle with time-varying dynamics manually. In the experiment the human pilot controlled basic first-order linear time-varying controlled elements in a compensatory closed-loop fashion for step and random inputs. The analysis of the experimental data has found that for the step input case the human pilot can be modeled as a stationary proportional-plus-integral controller, whereas for the random input case a stationary model with constant coefficients cannot describe the pilot's control behavior sufficiently well. A nonstationary time series analysis suggests that the human pilot's control behavior for the random input case is also time-varying; adapting his characteristics to the changing controlled element dynamics in such a manner that the error and output spectral characteristics do not make appreciable changes with respect to time.