Performance Analysis of LQG Controller for Stabilizing a Two wheeled Self-balancing Robot

Abstract

The stabilization of two-wheeled self-balancing robots (TWSBRs) is a critical challenge due to their inherent instability and complex dynamics. The concept of an inverted pendulum serves as a foundational model for engineers seeking to master their balance, ensuring the stability of these robots is paramount. Employing an adaptive intelligent control strategy, informed by the inverted pendulum principle, proves particularly effective when the robot initiates movement from varying initial tilt angles Comparative performance analysis of the linear quadratic gaussian (LQG) controller against the linear quadratic regulator (LQR) and its optimized variants, LQR-particle swarm optimization (LQR-PSO) and LQR-flower pollination algorithm (LQR-FPA), for stabilizing TWSBRs across various pitch reference angles. The analysis reveals that the LQG controller consistently demonstrates superior performance in minimizing both peak pitch deviation and peak wheel speed. For each pitch reference angle, LQG to achieve a peak pitch deviation of 0.035 degrees and a peak wheel speed of 0.045 degrees/second. Two-wheeled self-balancing robots, frequently modeled as inverted pendulums, represent a significant area of focus in robotics research. Their versatility spans applications from personal mobility to automated operations, and their inherent ability to maintain upright stability across diverse terrains renders them invaluable in various industries. Generally performance analysis of overall response self-balancing robot with LQR, LQR-PSO, LQR-FPA, and LQG at different reference angle 15°, 43° and 65°, that gives smaller peak pitch deviation and peak tracking error with LQR, LQR-PSO, LQR-FPA, LQG which is less than 1°. Design an optimal adaptive controller for self-balancing robots in the otherwise unstable vertical upright reference position. To accurately ascertain the robot lean angle and mitigate measurement uncertainties, Kalman filters are integrated. Simulation outcomes demonstrate the efficacy of this integrated approach in significantly bolstering robot stability, paving the way for more dependable self-balancing systems within the autonomous robotics domain.