Modified Integral Sliding Mode Controller Design based Neural Network and Optimization Algorithms for Two Wheeled Self Balancing Robot

Abstract

Two-wheeled Self-balancing (TWSB) mobile robot is considered to be highly nonlinear and unstable dynamic system. Unstable means that the robot is free to advance forward or backward without any forces applied. It must, therefore, be controlled. The purpose of this work is to design an intelligent nonlinear Modified Integral Sliding Mode Controller (MISMC) based on simple Adaline neural network for balancing a two-wheeled self-balancing mobile robot, in addition to improve the performance of this robot in tracking the desired trajectory. The simple Adaline neural network is used to enhance the performance of the conventional Integral Sliding Mode Controller (ISMC) which is an effective and powerful technique because it has a high performance. Also, in this work, a Modified Particle Swarm Optimization (MPSO) and Modified Cuckoo Search (MCS) algorithms have been proposed to find and tune the best MISMC parameters and hence enhance the performance characteristics of the robot system by reducing the processing time as well as improving the response accuracy through minimizing the tracking error of the mobile robot. The Integral Square Error (ISE) method has been used as a performance index for the two algorithms (MPSO, MCS) to measure the performance of the proposed controller. Numerical simulations show the efficiency of the suggested controller by handling the balance and tracking problems of the two-wheeled self-balancing mobile robot.