Data-driven modeling and adaptive event-triggered secure control for autonomous vehicles subject to sensor attacks.

Abstract

This paper is concerned with data-driven model identification and adaptive event-triggered secure control of autonomous vehicles subject to sensor attacks. Firstly, the lateral dynamical model of autonomous vehicles is identified from data by exploiting the dynamic mode decomposition (DMD) approach and the sensor attacks are considered based on the established model. Then, an adaptive event-triggered scheme is well developed to balance the communication efficiency and control performance. Thus, the sliding-mode-like control scheme is utilized to counteract the sensor attacks. The stability analysis and stabilization design are derived using Lyapunov theory and linear matrix inequalities technique. There are three advantages of the proposed control scheme: a) DMD overcomes modeling difficulties, b) event-triggered threshold is adaptively regulated by the feedback measurement and c) the sensor attacks can be actively mitigated. At last, several comparison examples show the effectiveness of the proposed data-driven secure control scheme.