Robots moving in formation by using neural network and radial basis functions

Abstract

Vision-based moving in formation by four mobile robots is presented. One robot who is a leader and goes first provides moving plans to the other robots who follow the leading robot. These robots move not only in a single line, but also triangular or diamond formation. Each robot detects the other robots by means of color image classification using a three-layer neural network. In motion control, a radial basis function (RBF) network approximated by learning is used. In addition, hardware implementations and the results of a demonstration of how multiple mobile robots move in several formations are described.<>