Utilizing Internal State in Multi-Robot Coordination Tasks

The success of a task-achieving multi-robot system (MRS) depends on effective coordination mechanisms to mediate the robots’ interactions in such a way that a given task is achieved. In the MRS community, many elegant coordination mechanisms have been empirically demonstrated. However, there is a lack of systematic procedures for the synthesis of coordinated MRS. In this paper, we address this issue by presenting a principled framework suitable for describing and reasoning about the intertwined entities involved in any task-achieving MRS – the task environment, task definition, and the capabilities of the robots themselves. Using this framework, we present a systematic procedure by which to synthesize controllers for robots in a MRS such that a given sequential task is correctly executed. The MRS is composed of homogeneous robots capable of maintaining internal state but not capable of direct inter-robot communication. This systematic approach to the synthesis of coordinated MRS permits formal identification of the benefits and limitations of MRS composed of robots maintaining internal state and when other types of controllers may become necessary, such as those using communication. The most relevant related work includes work on finding optimal policies in partially observable stochastic domains (?). In MRS, the study of information invariants (?) addresses the problem of determining information requirements for performing robot tasks. The work presented here builds upon our previous results on automated synthesis of controllers using internal state (?) by addressing the issue of uncertainty in sensing and action. Our work presented in (?) presents results on the synthesis of stateless controllers which are communicative but stateless. In the multi-agent systems community, the issues addressed in (?) and (?) are related to ours in that they are systematically studying the role different control characteristics play in coordinated systems.