ScholarWorks@숙명여자대학교

초록

With the increasing integration of advanced driver assistance systems (ADAS) in modern vehicles, intelligent regenerative braking systems (IRBS) have emerged as an effective solution for energy-efficient car-following control. However, current IRBS implementations lack personalization and require drivers to select fixed braking levels, often causing discomfort by failing to reflect individual preferences. Existing personalization techniques rely on deep learning or offline-trained models, which are computationally intensive and unsuitable for real-time deployment on vehicle electronic control units (ECUs). To address these limitations, this study proposes a computationally efficient parameter adaptation algorithm based on the extended Kalman filter (EKF), enabling real-time personalized car-following control in IRBS. The proposed method adopts a dynamic relative speed (DRS) spacing policy to capture individual driving characteristics. A reference data generator generates real-time reference data during specific driver interventions, allowing the EKF algorithm to adapt control parameters accordingly. Simulation tests using a driver model demonstrated a significant reduction in driver interventions and maintained inter-vehicle distances aligned with the driver's preferred distance. Vehicle tests further confirmed the algorithm's real-time performance and low computational load, affirming its suitability for large-scale production. This study presents a practical and scalable method for personalizing IRBS-based car-following control without complex learning architectures, thereby enhancing driving comfort and expanding applicability to ADAS technologies such as adaptive cruise control (ACC).

키워드