A Lifetime Management Framework for SOTIF-Oriented Vehicle Dynamics Control: A Data-driven Approach

Jicheng Chen,Jingguo Liang,Siyu Teng,Hui Zhang,Fei-Yue Wang
DOI: https://doi.org/10.1109/tiv.2023.3341144
IF: 8.2
2023-01-01
IEEE Transactions on Intelligent Vehicles
Abstract:Current safety of the intended functionality (SOTIF) studies are focused on perception limitations, decision uncertainties, and human-machine interaction, while it tends to overlook the impact of the control system on SOTIF. This letter seeks to address this gap by establishing a lifetime management framework for vehicle dynamics control (VDC) systems. At first, we provide an introduction to the fundamental composition and control architecture of a VDC system. Furthermore, we identify representative hazard scenarios, trigger conditions and performance deficiencies in VDC. Building upon data-driven fault-tolerant control theory, a comprehensive framework consisting of adaptive filter, iterative learning, fault diagnosis and identification, and fault-tolerant control modules is presented. The proposed framework is specifically designed to address trigger conditions such as driving environment variations, integration of multiple control systems, performance degradation due to components aging and maintenance. The framework features online driving state supervision, driving risk measurement, and intervention for mitigation of the SOTIF issues. As a summary of a part of Distributed/Decentralized Hybrid Workshop on Sustainability for Transportation and Logistics (DHW-STL), the letter reports the development of an intelligent vehicle lifetime management framework with a strong focus on SOTIF issues.
What problem does this paper attempt to address?