New platooning control estimates powertrain constants without persistent excitation

A team of researchers from Southeast University and related institutions has published a new adaptive longitudinal platooning method that solves a long-standing estimation problem: correctly identifying powertrain time constants in real time. The paper, titled "Correct Online Estimation of the Powertrain Time Constants in Adaptive Vehicular Platooning," appears in arXiv (2608.06835) and the IEEE Transactions on Systems, Man, and Cybernetics: Systems. The authors—Qiuhao Wen, Simone Baldi, Jiwei Wang, Wenwu Yu, and Di Liu—modified the composite adaptive control framework to handle uncertainty in the input matrix, which previously forced either impractical persistence-of-excitation requirements or an assumed-known input matrix.

The new protocol guarantees that estimated time constants converge to their true values under an extremely relaxed condition: the derivative of the vehicle's acceleration only needs to be nonzero over a possibly short transient. This is a major departure from classic adaptive control, which typically demands persistent excitation for parameter convergence. The design also eliminates the need to measure acceleration derivatives directly or collect past data, making it far more practical for real platooning systems. Comparative tests against state-of-the-art solutions demonstrated clear advantages, and CarSim-based platooning experiments confirmed the robustness of the approach under realistic driving scenarios. For autonomous truck convoys and connected vehicle fleets, this could mean more accurate powertrain models, tighter spacing, and better fuel efficiency without complex excitation maneuvers.