Development of Rear Toe Angle Kinematic and Compliance Steer Disturbance Compensation System Using Hybrid Estimation Model

This paper proposes and validates a hybrid estimation model to compensate for kinematic and compliance (K&C) steering disturbances in automotive rear-wheel steering (RWS) systems. The model integrates a regression-based kinematic model with a physics-informed neural network (PINN) that learns the prediction residuals to compensate for nonlinear compliance effects. The performance evaluation compared the lateral acceleration between the stand-alone kinematic and hybrid models. In double-lane-change simulations at 130 km/h, the hybrid model achieved an average 27.5% reduction in mean squared error (MSE) and 7.8% improvement in the coefficient of determination (R²) compared to the kinematic model. These results suggest that the proposed model can significantly improve disturbance compensation in practical RWS systems and serve as an effective foundation for precise vehicle dynamics control.