Residual Learning-Based Control of Vehicle Platoons with $\ell_2$ Stability Guarantees via Recurrent Equilibrium Networks

2026-09-27

AI Summary

This study addresses the control challenges of heterogeneous vehicle platoons subject to parametric uncertainties and external disturbances by proposing a robust control framework based on residual learning. The method designs a nominal controller via linear matrix inequalities (LMIs) and introduces a constrained Recurrent Equilibrium Network (REN) satisfying small-gain conditions for residual compensation, combined with a disturbance observer to ensure local closed-loop and string stability. By offline training the REN to augment the nominal control, experimental results demonstrate that the proposed framework significantly reduces spacing and velocity tracking errors, outperforming conventional nominal controllers.

Abstract

This paper proposes a residual learning-based control framework for heterogeneous vehicle platoons subject to parametric uncertainty and external disturbances. A nominal controller designed via Linear Matrix Inequalities (LMIs), along with disturbance-observer compensation, is enhanced by a Recurrent Equilibrium Network (REN) trained offline using stored trajectories and nominal-model prediction errors. The REN is constrained to satisfy a prescribed $\ell_2$-gain bound, enabling sufficient small-gain conditions for local closed-loop stability and disturbance string stability. Experiments demonstrate reduced spacing and velocity errors relative to the nominal controller.

Problem

• vehicle platoons
• parametric uncertainty
• external disturbances
• string stability
• heterogeneous systems

Innovation

• Residual Learning
• Recurrent Equilibrium Networks
• Vehicle Platoons
• String Stability
• Linear Matrix Inequalities