Advances in Neural Networks - ISNN 2006

Abstract

An important application of mobile robots is searching a region to locate the origin of a specific phenomenon. A variety of optimization algorithms can be employed to locate the target source, which has the maximum intensity of the distribution of some detected function. We propose a neural network based dual heuristic programming (DHP) algorithm to solve the collective robotic search problem. Experiments were carried out to investigate the effect of noise and the number of robots on the task performance, as well as the expenses. The experimental results were compared with those of stochastic optimization algorithm. It showed that the performance of the dual heuristic programming (DHP) is better than the stochastic optimization method. © Springer-Verlag Berlin Heidelberg 2006.