Design of RTD-PID optimized neural networks controller for non-holonomic wheeled mobile robot

Abstract

In fact, many efforts have been devoted to the tracking control of the two-wheeled and non-holonomic mobile robots and many types of controllers were applied to overcome trajectory tracking problems. Some of them have been investigated based on conventional methods using PID control,2,3 robust control,4,5 sliding mode control,6–8 adaptive control.9,10 The others are based on artificial intelligence using fuzzy control11–13 and neural control.14–16 In fact, neural networks are recommended for intelligent control of nonlinear dynamic systems. Principally, this is due to two important properties of neural networks: their ability to learn, and their good performance for optimization. Nowadays, much attention is devoted to the use of neural networkbased control of mobile robots for trajectory following. The principle of the method is based on a multilayer feed-forward neural networks with back-propagation learning algorithm, and it has been shown that only one hidden layer can be largely sufficient to approximate any continuous functions. In,3 a PID-based neural network is developed for velocity and orientation, tracking control of a non-holonomic mobile robot that is appropriate for a kind of plant with nonlinearity uncertainties and disturbances.