A unified adaptive control formulation for safe and high-performance ground vehicle lateral guidance

Abstract

In this paper, we propose a unified adaptive control formulation to improve the safety and performance of ground vehicle lateral guidance systems. This approach uniquely combines two features: prescribed performance control and hyperbolic sine-based attracting-manifold adaptation. The formulation offers two major benefits: first, prescribed performance control guarantees the forward invariance of tracking errors within predefined funnels, ensuring safety; second, the hyperbolic sine-based attracting-manifold adaptation enhances closed-loop performance by actively attenuating perturbations from parametric uncertainties and achieving adaptation without regret. By orchestrating these two mechanisms, the proposed approach outperforms designs that rely on either technique alone. Developed in a general framework, the formulation is applicable to a broad class of second-order control-affine systems in controllable canonical form. We demonstrate its effectiveness by applying it to ground vehicle lateral guidance and validate its performance through high-fidelity simulation using the CarSim-Simulink joint platform. Notably, the proposed controller achieved a preview tracking error convergence time of approximately 2 s—50% faster than baseline methods—and reduced peak lateral deviation by up to 66% during sharp lane-change scenarios, highlighting its practical benefits in both safety and performance.