Modelling and Control of a Self-Balancing Robot Using Sensor Fusion and LQR Controller

This paper illustrates a humanlike self-balancing robot's control and sensor fusion with an inherently imbalanced dynamical model. Using a state space model, a two-wheeled self-balancing robot (TWSBR) is modelled as an inverted pendulum. A comparative study of the system performance is conducted, with different control techniques, namely, PD, PID, and Linear Quadratic Regulator (LQR). The system status is monitored and fed back to the controller using an internal measurement unit (IMU) with data filtering through sensor fusion. The controlled system is simulated using MATLAB/ Simulink and Simscape. The quantitative performance of different controllers is calculated using seven different indices: maximum overshooting, settling time, rise time, root-mean-square error (RMSE), steady-state error, controller effort, and max control action. The results proved that the LQR gives the best performance with the lowest RMSE and reasonable rise, settling time, energy consumption, and saturation level; therefore, it is the most suited for our system requirements. © 2022 IEEE.