This study proposes a method for technical improvements to the automated vehicle dynamic performance evaluation system suggested in previous research. This system is an automated exploration system for feasible regions in the design space of Electrical control unit (ECU) parameters, using Bayesian active learning and X In the Loop Simulation. We introduce enhancements to the previous exploration method by adapting the acquisition function according to the exploration status. The purpose of the proposal method is to improve multi-objective exploration accuracy and evaluation efficiency compared to the previous method. To demonstrate the effectiveness of the proposal method, we conducted a comparison with the previous method using the exploration of feasible regions for ECU parameters that satisfy drivability performance targets as a case study. As a result, the proposal method demonstrated superior performance in the automated exploration of feasible regions by covering a broader range with fewer evaluations. The results of this study are expected to serve as a foundation for the automated design of ECU parameters in the development of vehicle dynamic performance.
YAMAMOTO et al. (Wed,) studied this question.