模型预测控制
工程类
功能(生物学)
控制(管理)
控制系统
车辆安全
控制工程
控制理论(社会学)
计算机科学
汽车工程
人工智能
进化生物学
生物
电气工程
作者
Seokju Hong,Yunhyoung Hwang
标识
DOI:10.1080/00423114.2025.2544993
摘要
The lane-keeping assistance system (LKAS) is key for safety among advanced driver assistance systems (ADAS). It prevents unintentional lane departures and is the standard in most new vehicles. Importantly, its commercial value necessitates minimising false positives resulting from unnecessary early interventions and complying with LKAS regulatory requirements. Therefore, this study proposes a controller that combines model predictive control (MPC) with a control barrier function (CBF), to strictly satisfy regulation-related constraints, even under delayed intervention. The controller was designed to intervene as late as possible for improved commercial value. However, such delayed interventions can reduce the size of feasible sets, interfering with feasibility. Thus, to ensure feasibility, the extended class K∞ function of the CBF constraint was optimised. The proposed controller was evaluated based on the European New Car Assessment Programme (Euro NCAP) for lane support systems. Experiments were conducted using MATLAB/Simulink and CarMaker, delaying controller intervention to evaluate the commercial value of the controller. To further validate the robustness of the controller, additional experiments were conducted under disturbances, such as crosswinds and additional loads. The results demonstrate that the proposed controller outperforms comparative controllers in terms of satisfying regulation-related constraints under various driving conditions.
科研通智能强力驱动
Strongly Powered by AbleSci AI