Characteristics of object-searching and object-fetching behaviors of multi-robot system using local communication

Abstract

We introduce a methodology using emergent intelligence to generate some of useful behaviors in a multi-robot system consisting of robot modules equipped with a simple function and intelligence mechanism. The generated abilities of a multi-robot system are as follows: one is the ability to explore their unknown area; the other is the ability to return to their original position from any position in their work area. We have demonstrated these behaviors by using our manufactured robots and by computer simulation. The characteristics of two basic abilities in the emergent intelligence are investigated. One ability, which is the searching ability or space-coverage ability, is generated by random-walk-like moving that emerges by dynamic interaction in each robot when there is an increase in the population size of the multi-robot system. The other ability, which is the returning ability, is increased effectively using a vector-field generated by the signal flow via a communication network. The paper introduces two approaches to building a communication network. One is to use a static signal-transmission robot and the other is to use a moving signal-transmission robot. When any one of individual robots finds a certain area and then returns to its original area, the total ability obtained by the method using the moving signal transmission robot is superior to that by the method using a static signal-transmission robot. On the other hand, the probability evaluated as the ratio of the searching frequency to the returning frequency for certain time duration, is almost the same between the two approaches. We conclude that the high probability of returning to the original position after searching is a fundamental ability utilizing a vector-field via a communication network when compared to the return probability without utilizing a communication network.