Managing teams of unmanned vehicles is currently time-consuming and labor intensive. There needs to be a way to control multiple UAV teams with minimal human oversight. The proposed innovation builds on and combines several technologies we have developed to create an architecture and set of software methods that will achieve this goal, significantly advancing the state of the art. The proposed innovations are based on our NASA-funded Aurora planning, resource allocation, and scheduling framework, which has proved optimal in many, many diverse domains, including UAV scheduling; a Probabilistic RoadMap Planner (PRMP) to plan detailed real-time UAV routes to rapidly satisfy and optimize a large number of simultaneous constraints and objectives; the asynchronous consensus-based bundle algorithm (ACBBA) for UAV-to-UAV task negotiation; and the concept of a play (from sports) represented using behavior transition networks (BTNs). The ultimate goal of this proposed effort is to allow intelligent UAV team coordination and control in an intelligent, predictable, and robust way, with little cognitive load on the human users. This will require intelligent real-time planning, role allocation, negotiation, and detailed path planning and, when communication is not possible, autonomous, intelligent, adaptive behavior by the UAVs. In Phase I, we will develop the required AI techniques to automate all aspects of intelligently executing, recommending, and/or automatically selecting appropriate plays, robustly assigning roles and planning routes, and adaptively executing each role, robustly and predictably in environments with varying levels of uncertainty. We will design the ultimate system and, to absolutely prove its feasibility, prototype all aspects of it in Phase I on *actual, physical UAVs*.