Modelling and control of a flexible two-wheeled self-balancing mobile robot

Abstract

This paper presents a dynamic control model for a flexible two-wheeled self-balancing robot based on Lagrange’s equation and the theory of dynamics and mechanics. One new aspect, introduced in this paper, is the modelling of the human lumbar – a spring is added to the robot to imitate lumbar flexibility and curvature. The validity of the system modelling and controller design is verified through simulation, experimental results and the implementation of the robot. Two methods were used to control the robot’s posture: a Discrete Hopfield Neural Network (DHNN) and a Boltzmann machine; the results are compared and then an improved Boltzmann machine is described.