Reliability- and Sensitivity-Guided Co-Estimation of Vehicle States, Tire Cornering Stiffness, and Tire-Road Adhesion Coefficient

Reliable estimation of vehicle states, tire cornering stiffness, and the tire-road adhesion coefficient are essential for vehicle lateral stability and intelligent chassis control. Under low adhesion, nonlinear tire operation, and weak excitation, lateral-force residuals are jointly affected by cornering-stiffness variation, adhesion-coefficient variation, and tire-force saturation, which may cause erroneous parameter adaptation. This paper proposes a reliability- and sensitivity-guided co-estimation method for vehicle states, tire cornering stiffness, and the tire-road adhesion coefficient. A hierarchical framework is developed based on a planar 3-DOF vehicle model and a Fiala-type nonlinear tire model. Front- and rear-axle lateral-force pseudo-measurements are reconstructed from lateral acceleration and yaw angular acceleration, without requiring additional tire-force sensors. Parameter-update reliability is evaluated by considering lateral excitation, longitudinal slip, adhesion utilization, and normalized lateral-force residual consistency. Normalized lateral-force sensitivities are then used to allocate the residual between the cornering-stiffness and adhesion-coefficient update channels. CarSim/Simulink co-simulations under high-, intermediate-, and low-adhesion double-lane-change maneuvers demonstrate that the proposed method improves sideslip-angle and lateral-velocity estimation accuracy, suppresses erroneous cornering-stiffness adaptation, and provides more stable estimates of the tire-road adhesion coefficient.