Neural network self-tuning PID based navigation control of autonomous unicycle-like mobile robot in industry 4.0

Abstract

In the recent years, mobile robots, one of the technologies under the “Industry 4.0” concept, have been used in a wide range of industry sectors, including manufacturing and production, agriculture, healthcare, etc. One of the applications of a mobile robot is transportation to deliver things from one place to another, following the planned trajectory. The conventional way of controlling the trajectory tracking of a mobile robot is by using the classical PID control schemes. However, it has been found that the performance was not very satisfying because PID controllers have a weak adaptability to the mobile robot dynamic system which consists of nonlinearity and uncertainty that varies with time. In order to achieve adaptive controller, this study proposes a Neural Network (NN) self-tuning PID based navigation control which is capable to perform on-line tuning of the PID parameters to meet the desired control performance and stability during operation. In this work, MATLAB-Simulink software is used to simulate the dynamic model of a unicycle-like mobile robot. PID controllers which are tuned with the Trial & Error method is firstly used to control the trajectory tracking of the mobile robot. Then, the same dynamic model is controlled by using the proposed NN self-tuning PID controllers. The simulation results obtained from both simulations are compared from the aspect of the distance error and energy consumption by calculating the IAE index and kinetic energy index, and the results show the capability of the NN self-tuning PID controllers to perform better than a PID controller in a non-linear system.