SOTIF-Oriented Distributed Control for Connected Autonomous Vehicles Platoons
针对网联自动驾驶车辆队列，提出一种基于分布式观测器的鲁棒弹性控制协议，在对抗条件下维持预期功能安全性能，并通过仿真验证了有效性。
This article extends the problem of safety of the intended functionality (SOTIF) to the platooning of connected autonomous vehicles (CAVs). In such multivehicle systems, SOTIF objectives are particularly challenging, since new issues arise due to the presence of interactions that propagate and intensify the effects and the risk connected to faults/failures/uncertainties, triggered by any platoon member and external conditions. To tackle and solve the SOTIF-oriented platoon control problem, we propose a robust and resilient decentralized control protocol, based on distributed observer which exploits the information shared via the Vehicle-to-Everything (V2X) communication network. The approach allows maintaining acceptable SOTIF performance in adversarial conditions, mitigating the effects of the unreasonable risks originated from sensors/actuators performance limitations/faults and disturbances. The uniformly ultimate bounded (UUB) stability and the input-to-state string stability (ISSS) of the platoon dynamics are analytically proved via the Lyapunov theory. Feasible linear matrix inequalities (LMIs) are obtained as stability conditions, whose solution allows adapting the distributed control/observer gains according to leader-tracking performance and robustness margins. A detailed analysis via the vehicular co-simulation platform mixed traffic simulator (MiTraS) corroborates the theoretical derivations and shows the achievement of the SOTIF platoon goals for a set of nontrivial and uncertain driving scenarios.