Group behaviour control on dynamically reconfigurable robotic system

Recently, distributed autonomous robotic system (DARS) has been attracting many researchers' interests. As one of the model of DARS, we proposed cellu lar robotic system (CEBOT). Like an organism of a living thing, the CEBOT i s composed of a number of heterogeneous robotic units called 'cells'. Based on the concept of CEBOT, we have studied a flexible and adaptive multiple robot system by using a production transit model. In this model, we assume a number of mobile robots that have a few kinds of manipulators and variati on of transportation ability according to the combination of the different manipulation functions. We suppose that by changing its combination appropr iately against environment changes, the adaptability of the system can be i mproved. At first, we address the combination making and show the advantage of the proposed system. Generally, in a swarm-type robot system, we can fi nd some sort of collective phenomena that are not defined in each robot's b ehavior decision process. Oscillation is one for the typical collective pro perties emerged in DARS. In our model, the occurrence oscillation affects t he performance of transportation. Therefore, we need a way of controlling s uch collective property as well as the appropriate combination making. In t his paper, we show the advantage of the dynamically reconfigurable robotic system and a control method of collective property through a computation si mulation.