Title:Robust Data-Driven Predictive Control for Mixed Platoons under Noise and Attacks

Abstract:Controlling mixed platoons, which consist of both connected and automated vehicles (CAVs) and human-driven vehicles (HDVs), poses significant challenges due to the uncertain and unknown human driving behaviors. Data-driven control methods offer promising solutions by leveraging available trajectory data, but their performance can be compromised by process noise and adversarial attacks. To address this issue, this paper proposes a Robust Data-EnablEd Predictive Leading Cruise Control (RDeeP-LCC) framework based on data-driven reachability analysis. The framework over-approximates system dynamics under noise and attack using a matrix zonotope set derived from data, and develops a stabilizing feedback control law. By decoupling the mixed platoon system into nominal and error components, we employ data-driven reachability sets to recursively compute error reachable sets that account for noise and attacks, and obtain tightened safety constraints of the nominal system. This leads to a robust data-driven predictive control framework, solved in a tube-based control manner. Numerical simulations and human-in-the-loop experiments validate that the RDeeP-LCC method significantly enhances the robustness of mixed platoons, improving mixed traffic stability and safety against practical noise and attacks.