TY - JOUR
T1 - Distributed State Estimation and Fault-Resilient Strategy for Safe Autonomous Driving under Multisensor Faults
AU - Lu, Yu
AU - Li, Guoqiang
AU - Li, Zhenyang
AU - Wang, Zhenpo
N1 - Publisher Copyright:
© 2007-2012 IEEE.
PY - 2026
Y1 - 2026
N2 - Accurate vehicle state information is critical for safe and stable autonomous driving, yet sensor faults can distort measurements, degrade control performance, and even trigger driving failures. To address this issue, this article proposes an adaptive resilient fault-tolerance framework based on multiinformation fusion for vehicle motion control under multisensor fault conditions. A distributed state estimation scheme first fuses multisensor data to improve estimation accuracy. Then, a distributed fault detection and diagnosis module identifies faulty sensors, isolates faulty sensor channels, and evaluates fault severity. After fault isolation, a rule-based resilient reconstruction strategy excludes corrupted sensor channels and reconstructs reliable states by fusing the remaining healthy observer outputs. Based on the reconstructed states, an SMPC controller achieves constrained optimal path tracking. CarSim–Simulink cosimulation results demonstrate that the proposed framework effectively improves state estimation, fault diagnosis, and control performance, thereby enhancing driving stability and safety under complex sensor fault conditions.
AB - Accurate vehicle state information is critical for safe and stable autonomous driving, yet sensor faults can distort measurements, degrade control performance, and even trigger driving failures. To address this issue, this article proposes an adaptive resilient fault-tolerance framework based on multiinformation fusion for vehicle motion control under multisensor fault conditions. A distributed state estimation scheme first fuses multisensor data to improve estimation accuracy. Then, a distributed fault detection and diagnosis module identifies faulty sensors, isolates faulty sensor channels, and evaluates fault severity. After fault isolation, a rule-based resilient reconstruction strategy excludes corrupted sensor channels and reconstructs reliable states by fusing the remaining healthy observer outputs. Based on the reconstructed states, an SMPC controller achieves constrained optimal path tracking. CarSim–Simulink cosimulation results demonstrate that the proposed framework effectively improves state estimation, fault diagnosis, and control performance, thereby enhancing driving stability and safety under complex sensor fault conditions.
KW - Fault detection and diagnosis
KW - multiinformation fusion
KW - resilient fault-tolerance strategy
KW - sensor fault
KW - state estimation
UR - https://www.scopus.com/pages/publications/105050196749
U2 - 10.1109/JSYST.2026.3720665
DO - 10.1109/JSYST.2026.3720665
M3 - Article
AN - SCOPUS:105050196749
SN - 1932-8184
JO - IEEE Systems Journal
JF - IEEE Systems Journal
ER -