NARXモデルを用いた，車両の動特性と運転しやすさの関係に関する一考察

抄録

This paper describes an investigation on the nonlinear filter that can realize its gain and phase characteristics independently. In the driver-vehicle closed loop system, it is reported that there is an optimal value in vehicle phase delay, i.e., excessive large or small delay deteriorates vehicle maneuverability. However, there is a strict relationship between the gain and phase delay in linear system. It is difficult to evaluate the gain and phase delay of vehicle response independently. In this investigation, we tried to design a non-linear filter that can set the relationship between gain and phase delay arbitrarily by using neural network. We investigate the effect of the gain and phase delay corresponding to the steering angle input on the ease of driving by using driving simulator, and evaluate the control performance on the lane change task. We verified that the ease of driving is improved by reducing a phase delay and decreasing a high frequency gain.