Codesign of High-Order Polynomial-Driven Observer and Safety-Oriented Gain-Scheduling Scheme for Cloud-Assisted Shared Steering Systems

As intelligent connected technology gradually integrates into conditional driving automation, human–machine shared driving vehicles face severe cybersecurity threats. To mitigate the path-tracking performance degradation of shared driving vehicles caused by malicious attacks and risky driving, this article explores safe-oriented gain-scheduling modes by utilizing the driving risks assessment and high-order polynomial-driven observers. First, this article combines interval type-2 fuzzy inference systems (IT2FIS) with multidimensional driving risks assessment to design a lightweight dynamic human–machine authority allocation strategy for the driver-vehicle-path system. Second, the multimode nonparallel high-order free-weighting matrix-scheduling (MNHFM) controller is developed to resolve the superimposed difficulties caused by state loss and nonlinear parameter changes through flexibly leveraging fuzzy membership relationships. Meanwhile, polynomial-dependent time-varying reciprocal matrices are introduced to simplify the stability analysis under randomly triggered network attacks. Finally, driver-in-the-loop (DIL) experiments verify the progressiveness of the proposed MNHFM controller. The results show that the proposed control scheme can adaptively optimize the human–machine authority according to driving risks and has advantages in improving vehicle performance and ensuring driving safety.