Stabilization Control of Auto-balancing Bicycle subject to disturbances

In recent years, underactuated systems have been widely employed in engineering due to their simple structure, high performance, and low cost. This paper deals with the robust stabilization problems of a riderless bicycle at low speed. To facilitate the analysis, non-singular coordinate transformation is applied to dynamic model of the bicycle and an effective control method is proposed. By dividing the new dynamic model into two separate subsystems, the control objective is achieved by introducing a new virtual control input and disturbance rejection controller. The linear extended state observer is implemented to approximate and compensate for the lumped disturbances, including parametric uncertainties, internal non-linearity and external disturbances. Simulation and experimental results on an auto-balancing bicycle show the satisfying performances of the proposed method for the stabilization.