Tracking Control of a Nonholonomic Mobile Robot Using Neural Network

Tracking Control of a Nonholonomic Mobile Robot Using Neural Network

Şahin Yildirim (Erciyes University, Turkey) and Sertaç Savaş (Erciyes University, Turkey)
Copyright: © 2015 |Pages: 31
DOI: 10.4018/978-1-4666-7387-8.ch020


The goal of this chapter is to enable a nonholonomic mobile robot to track a specified trajectory with minimum tracking error. Towards that end, an adaptive P controller is designed whose gain parameters are tuned by using two feed-forward neural networks. Back-propagation algorithm is chosen for online learning process and posture-tracking errors are considered as error values for adjusting weights of neural networks. The tracking performance of the controller is illustrated for different trajectories with computer simulation using Matlab/Simulink. In addition, open-loop response of an experimental mobile robot is investigated for these different trajectories. Finally, the performance of the proposed controller is compared to a standard PID controller. The simulation results show that “adaptive P controller using neural networks” has superior tracking performance at adapting large disturbances for the mobile robot.
Chapter Preview

1. Introduction

The inspiration of technology in the new mechatronics era is the design of human-oriented machines that implies intelligent and cooperative coexistence beyond time and physical constraints (Habib, 2007). So mechatronics and robotics constitute one of the keys to human social and economic development (Tzafestas, 2014a; Tzafestas, 2014b). The benefits of using autonomous and intelligent mobile robots as a mechatronic system in medical, assistive and service applications are numerous and their positive impact in modern society is steadily increasing, because these robots can perform various actions wisely and autonomously. A mobile robot is suitable for many types of applications that require high autonomous in a wide variety of fields, just like in the factory and industry, in the society (rescue, guidance, safety, hospital services, cleaning, entertainment and other services). So the interest of mobile robots in scientific studies has been increasing rapidly due to the width of the potential of applications.

If we consider the most important ability of mobile robots is autonomous mobility, it has understood that the performance of tracking prescribed trajectories is very important. The tracking control approaches for the mobile robots are mainly divided into six types (Ye, 2008):

  • 1.

    State feedback linearization;

  • 2.

    Sliding-mode control;

  • 3.

    Backstepping control;

  • 4.

    Computed torque;

  • 5.

    Adaptive control;

  • 6.

    İntelligent control.

Different control systems have been used in previous works about wheeled mobile robots. Haddad et al. (2007) have presented a random-profile approach to treat the optimal free-trajectory planning problem with various forms of optimization criteria involving travel time, efforts and power for nonholonomic wheeled mobile robots subjected to move in a constrained workspace. Also Huang (2009) has applied the potential field method in a dynamic environment where the target and the obstacles are moving. The method was applied for both path and speed planning, or the velocity planning, for a mobile robot to make the robot track the moving target while avoiding moving obstacles.

Martins et al. (2008) have proposed an adaptive controller to guide tracking a trajectory of an unicycle wheeled mobile robot. The parameters of the robot dynamics were updated on-line, because these parameters can vary, such as load transportation. Chen et al. (2009) have designed an adaptive sliding-mode dynamic controller for wheeled mobile robots to implement the trajectory-tracking mission by making the real velocity of the wheeled mobile robot reach the desired velocity command, although the system uncertainties and disturbances. Klancar and Skrjanc (2007) have applied a model-predictive trajectory-tracking control to a wheeled mobile robot. Linearized tracking-error dynamics was used to predict future system behaviour and a control law was derived from a quadratic cost function penalizing the system tracking error and the control effort.

Key Terms in this Chapter

Artificial Intelligence: Approach considering intuitive sense contrarily procedural computing.

Adaptive Control: Control approach whose control parameters tuned during control process.

Back-Propagation Algorithm: Learning algorithm for neural networks.

Posture Tracking Error: Error between desired and actual trajectory.

Tracking Performance: Performance criteria for mobile robots while driving on a trajectory.

Robotic Systems: Systems using robotic hardware to run applications automatically.

PID Control: Proportional/Integral/Derivative control algorithm.

Online Learning Algorithm: Learning algorithm for neural networks during control process.

Complete Chapter List

Search this Book: