Balancing a Stationary Bicycle With a Reaction Wheel

Abstract

This thesis presents the design, modeling, and experimental validation of a stabilization system for a stationary bicycle using a reaction wheel actuator. The bicycle is modeled as an inverted pendulum, and a state-space representation is derived by combining mechanical roll dynamics with the electrical dynamics of a DC motor. The model serves as the foundation for controller design and simulation. Control strategies are developed and evaluated in MATLAB, where system behavior is analyzed under varying initial conditions, disturbances, and time delays. The cascade controlled architecture was proven as the most optimal. Furthermore, the pole placement method is applied to obtain desired closed-loop dynamics, and the influence of key parameters on system stability is investigated. The control algorithm is implemented on an embedded platform and tested on a physical prototype. Sensor data from an inertial measurement unit is processed using a complementary filter to estimate the roll angle in real time. Experimental results show that the system can generate stabilizing torque through the reaction wheel, but performance is limited by actuator response and system delays. Modifications, including compensation for back electromotive force, improve responsiveness and stability. The results demonstrate the feasibility of stabilizing a stationary bicycle using a reaction wheel system, while highlighting practical limitations related to hardware constraints and control implementation.