Multi-Robot Collaborative Task Performance based on Evolution of Neural Controllers

 G. Capi

Department of Electrical and Electronic Systems Eng., Faculty of Engineering, University of Toyama, Toyama, 3190 Gofuku, Toyama,  Japan

Abstract
This paper presents a new method for multi-robot collaborative task performance based on evolution of neural controllers. In our method, a neural network is used as behaviour decision controller for the robots that have to transport an object from its initial to the final location. The robots have to learn a correlation between their position and orientation and the position and orientation of the object in order to determine the best action.  Evolution took place online using the e-Puck robots. The experimental results show that evolution of neural controllers can be applied effectively for multi robot collaborative task performance.

 Keywords: Multi-robot system, Neural Network, Genetic Algorithms.

(Full Paper, 1.31 MB)

 

@ARTICLE{P1130916724,

AUTHOR = {G. Capi},

TITLE = {Multi-Robot Collaborative Task Performance based on Evolution of Neural Controllers},

JOURNAL = {ICGST  International Journal on Automation, Robotics and Autonomous Systems},

YEAR = {2009},

VOLUME = {09},

ISSUE={I},

PAGES={11--16}

}

(Full Paper, 1.31 MB)