Unified Modeling and Motion Prediction of a Reconfigurable Wheel-Tracked Vehicle via Instantaneous Center of Rotation

A reconfigurable wheel-tracked (RWT) vehicle combines the complementary advantages of wheeled and tracked locomotion, offering enhanced mobility in complex environments. However, its kinematic and dynamic characteristics vary significantly across different operating modes, while slip and skid effects during steering further aggravate model mismatch, posing challenges for accurate motion prediction and control. To address these issues, this paper proposes a unified modeling framework for an RWT vehicle based on the instantaneous center of rotation (ICR). A mode-independent kinematic model parameterized by the ICR is first established to uniformly describe vehicle motion under four configurations: wheel, inward-tracked, outward-tracked, and full-tracked modes. A corresponding dynamic model suitable for distributed-drive RWT architectures is then derived. Experimental validation on a paved surface demonstrates that the proposed dynamic model achieves satisfactory torque prediction accuracy, with an overall mean absolute error of 8.95 N· m and a normalized mean absolute error of 19.83%. Furthermore, an unscented Kalman filter (UKF) is employed to estimate mode-dependent ICR parameters from experimental data, enabling implicit compensation for slip effects without resorting to complex terramechanics modeling. Compared with conventional slip-neglecting kinematic models, the proposed ICR-based model reduces trajectory prediction errors by an average of 63.13%, with a maximum improvement of 74.5% across different modes. The ICR parameters identified through this framework provide a foundation for future control system design. Overall, the proposed framework provides an effective and practical solution for unified modeling of reconfigurable wheel-tracked vehicles operating under multi-mode and highly nonlinear conditions.