Fish Swarmed Kalman Filter for State Observer Feedback of Two-Wheeled Mobile Robot Stabilization

Abstract

Over the past few decades, there have been significant technological advancements in the field of robots, particularly in the area of mobile robots. The performance standards of speed, accuracy, and stability have become key indicators of progress in robotic technology. Self-balancing robots are designed to maintain an upright position without toppling over. By continuously adjusting their center of mass, they can maintain stability even when disturbed by external forces. This research aims to achieving and maintaining balance is a complex task. Self-balancing robots must accurately sense their orientation, calculate corrective actions, and execute precise movements to stay upright. Eliminating disturbances and measurement noise in self-balancing robot can enhance the accuracy of their output. One common technique for achieving this is by using Kalman filters, which are effective in addressing non-stationary linear plants with unknown input signal strengths that can be optimized through filter poles and process covariances. Additionally, advanced Kalman filter methods have been developed to account for white measurement noise. In this research, state estimation was conducted using the Fish Swarm Optimization Algorithm (FSOA) to provide feedback to the controller to overcome the effects of disturbances and noise in the measurements through the designed filter. FSOA mimics the social interactions and coordinated movements observed in fish groups to solve optimization problems. FSOA is primarily used for optimization tasks where finding the global optimal solution is desired. The results show that the use of an optimized Kalman filter with FSOA on a two-wheeled mobile robot to handle system stability reduces noise values by 38.37%, and the system reaches a steady state value of 3.8 s with a steady error of 0.2%. In addition, by using the proposed method, filtering disturbances and measurement noise in self-balancing robot can help improve the accuracy of the self balancing robot’s output. System response becomes faster towards stability compared to other methods which are also applied to two-wheeled mobile robots.