Risk-aware model predictive control for autonomous vehicle platoons under uncertain cut-in scenarios based on Gaussian mixture models

This study investigates the safety challenges faced by connected and automated vehicle (CAV) platoons operating in mixed traffic environments with human-driven vehicles (HDVs). In heterogeneous traffic scenarios, HDVs may abruptly cut into CAV platoons due to lane-changing or overtaking maneuvers, which can disrupt platoon stability and trigger cascading braking responses, thereby degrading traffic efficiency and ride comfort. To address the uncertainty and risk induced by HDV cut-in behaviors, a risk-aware model predictive control (MPC) framework based on Gaussian mixture models (GMMs) is proposed, in conjunction with the alternating direction method of multipliers (ADMM) for multi-vehicle cooperative control. First, the uncertain motion of HDVs, including position and velocity variations, is probabilistically modeled using GMMs to capture the multimodal characteristics of cut-in behaviors. Then, a GMM-based risk metric is incorporated into the MPC objective function in the form of a soft constraint penalty, yielding a risk-aware optimization framework that statistically guides the controller away from high-risk operating regions. Finally, ADMM is employed to realize distributed velocity coordination between the lead vehicle and the following vehicles, thereby enhancing the overall cooperative performance of the platoon. Simulation results obtained from a co-simulation platform integrating PreScan, CarSim, and MATLAB/Simulink demonstrate that, under platoon speeds of 40 km/h, 50 km/h, and 60 km/h, the proposed method outperforms conventional MPC and LQR controllers in terms of lateral spacing maintenance, trajectory tracking accuracy, and acceleration smoothness, exhibiting superior robustness and safety in the presence of uncertain cut-in disturbances.