A neural network based controller for a wheel type mobile robot

Abstract The transportation using wheels is one of the most popular transportation mechanisms for mobile robots because of its high energy efficiency, simple mechanisms and well-investigated control systems. However, the wheel type mobile robots have the biggest weakness in the rough terrain movement. In the previous researches, a six-wheeled mobile robot employing a linkage mechanism, “Zaurus”, has been developed to extend maneuverability of the conventional wheel type mobile systems. The experiments to climb over single bumps with twice the height of diameter of the robot's wheel have been carried out and succeeded. In this paper, a neural network controller and PID controller are introduced as the control system, and their performances are compared by simulations. The neural network controller shows superior filtering and feature extraction capabilities.

[1]  Robert Ivlev,et al.  The Rocky 7 Mars rover prototype , 1996, Proceedings of IEEE/RSJ International Conference on Intelligent Robots and Systems. IROS '96.

[2]  R. Siegwart,et al.  An Innovative Space Rover with Extended Climbing Abilities , 2000 .

[3]  Robert Ivlev,et al.  The Rocky 7 rover: a Mars sciencecraft prototype , 1997, Proceedings of International Conference on Robotics and Automation.

[4]  Mitsuhisa Sato,et al.  A Neural Network Based Control System for a Mobile Robot Employing Link Mechanism , 2005, AIM 2005.

[5]  Yasuharu Kunii,et al.  Low Power Mobility System for Micro Planetary Rover Micro 5 , 1999 .

[6]  Yoshinori Sato,et al.  Mobility performance evaluation of planetary rover with similarity model experiment , 2004, IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA '04. 2004.