Modeling and Parameters Estimation of a Self-Balancing Two-Wheeled Vehicle

Abstract

The self-balancing two-wheeled vehicle is a practical realization of the well-established control benchmark of the inverted pendulum system. Despite the sophisticated dynamics and kinematics of the inverted pendulum system, it has acquired great interest in real-life applications starting from Segway and hoverboards to the self-balancing wheelchair. These applications benefit from the dynamical structure that provides high maneuverability within narrow spaces. However, the system complexity regarding the high nonlinearity and instability requires accurate models for model-based control of the balancing and motion planning control objectives. This work addresses the modeling and the parameters estimation of a lab-scale version of the self-balancing two-wheeled vehicle. First, a nonlinear dynamical model based on LaGrange kinematics is developed. Then, real-time datasets for closed-loop operations are acquired for offline optimization. Finally, an optimization problem is formulated and solved for the parameter estimation through a decoupled block-by-block approach. The obtained grey-box model is validated for reliable and accurate fitting of the real system. The obtained model was able to simulate the real-time collected data with reasonable meaningful estimated parameters.