I. Introduction
When dealing with heterogeneous multi-robot teams, two issues are particularly challenging: (1) determining how to share sensor and perceptual resources across heterogeneous team members, and (2) determining the appropriate teaming behaviors to accomplish a task when the definition of how to solve a task is dependent on the available collection of robots and their sensory, perceptual, and motor resources. In typical approaches to multi-robot teaming, the task tree describing the decomposition of the team task into subtasks is defined by the human designer in advance of the robot team performance. The robots then choose from one of the alternative task decomposition trees, followed by the use of an automated approach for task allocation to determine the dynamic mapping of subtasks to robots. In these typical approaches, the predefined task decomposition tree defines the available multi-robot task solutions in advance of the mission (i.e., the how). This paper describes a new methodology for automating the synthesis of multi-robot task solutions in a general way that addresses both of the above challenges.