Design of Online Fuzzy Tuning LQR Controller Applied to Rotary Single Inverted Pendulum: Experimental Validation

In the present work, a new online fuzzy tuning linear quadratic regulator (FT-LQR) is proposed for balancing and trajectory tracking of the rotary single inverted pendulum. The beauty of the proposed approach is that it tunes the LQR weighting matrices Q and R online by using a fuzzy controller. The idea of the proposed controller is inspired by the designer's choice of Q and R matrices, where according to the system response, the designer changes the values of the two matrices to achieve the best performance. However, it is not easy to define these two matrices in an integrated manner. Classic methods such as trial and error require intensive efforts, time-consuming and do not guarantee the desired performance. Therefore, this empirical knowledge was applied to a fuzzy controller to adjust the weighting matrices according to the required performance and depending on the system's state. The FT-LQR is designed through simulation, and further experimental validations using a laboratory prototype verify the correctness of the proposed approach. Moreover, it can ensure a faster response, good stability, and better robustness to external disturbances in the different operating conditions.

Automated summary

In this paper, an online fuzzy tuning linear quadratic regulator (FT-LQR) is proposed for the balancing and trajectory tracking of a rotary single inverted pendulum. The FT-LQR tunes the LQR weighting matrices online using a fuzzy controller, allowing for better performance and stability.