Real-time decentralized multi-actuator chassis control strategy for vehicle lateral stability based on nonlinear MPC

This paper proposes a novel Decentralized Multi-Actuator Chassis Control (DMCC) strategy for vehicle lateral stability, utilizing real-time Nonlinear Model Predictive Control (NMPC). The proposed hierarchical framework decomposes integrated chassis control into a vehicle-level lateral dynamics control and multiple structurally identical wheel-level control for tire force realization. This modular design offers exceptional configurability, adapting seamlessly to diverse electric vehicle chassis configurations ranging from basic chassis configuration to fully actuated platforms featuring four-wheel independent drive, steering, and braking. To ensure microsecond-level deterministic computation, the vehicle-level controller employs a linearized model within an analytical quadratic programming formulation. At the wheel level, a Piecewise Affine (PWA) tire model with precomputed region-specific Jacobians facilitates efficient real-time linearization. Furthermore, a modified Hildreth’s active-set solver, augmented with Cholesky factorization and physics-informed warm-start strategies, is implemented to exploit spatiotemporal correlations, thereby achieving rapid convergence. High-fidelity CarSim co-simulations of high-speed double lane change maneuvers under varying friction conditions validate the approach. Results demonstrate substantial improvements in yaw rate tracking and sideslip constraint adherence compared to both uncontrolled baselines and conventional integrated control strategies, confirming the effectiveness and scalability of the proposed architecture.