Defense Science Board Summer Study on Autonomy Autonomy could enable a much faster planning cycle through the use of integrated tools and models that can handle the allocation of large numbers of resources in complex sequences with high interdependencies. It will be essential in managing and optimizing the complexity of branching scenarios associated with what-if analyses and pre-planning for contingencies. And, during execution, it enables a running comparison of “plan” to “actual,” to assess whether large-scale replanning is merited as events unfold. In this sense, planning is no longer a phase that happens before execution. Instead it becomes a continuous, background process that assists the commander in redirecting assets when needed during execution. The tools to shorten the MAAP/ATO process by an order of magnitude do not yet exist. However, there is evidence of enabling progress in many underlying capabilities. For example, DARPA’s Distributed Battle Management program is developing automated decision aids for managing air-to-air and air-to-ground combat. Autonomy is required to handle the scale and complexity of the planning at speed. The study notes that a better planning tool on its own is not sufficient to shorten the planning timeline. That will also require changes in workflow, staffing, and other factors. Recommendation 26. The AFRL/RIS office should undertake a stretch problem to generate a new MAAP/ATO within one hour of target development. The cost is estimated at $25 million per year, with tests every 18 to 24 months until the objective is achieved. • This stretch problem differs from some others in that it is tied to a specific operational process. Thus, its form appears most like a standard program—but with some essential features to drive innovation from the commercial world. One critical element is the necessity for an open architecture, to enable participation by a variety of performers. An open architecture also supports the development of modular mission capabilities, so that the mission-planning scope can be expanded over the course of the program. The program manager should spend the first year of execution defining the open architecture to be used within the program. • Another essential feature is a program structured around a series of milestones that assign increasingly challenging time goals for ATO generation, e.g., 12 hours, six hours, one hour. The program manager should apportion an expanding scope of mission capabilities to each milestone. Once the architecture is defined, the government should open a competition to fund at least two competing systems integrators, each leading an innovative team of subcontractors that bring diverse and leading technology expertise. Periodically (e.g., every 18 months) the sponsor should run a head-to-head competition of performance against the current milestone goals. The winning team should be assigned a prize, possibly in the form of an award fee. Then each prime should be allowed to reselect a new team of subcontractors, based on performance to date and emerging technologies. • The program should carry out the assessments at interim milestones using a combination of live, virtual, and constructive facilities. The graduation exercise should be carried out at Red 96
Defense Science Board Summer Study on Autonomy Flag, to enforce performance assessment in a realistic combat environment that mimics the fully complexity of modern operations—fighter interdiction, attack/strike, air superiority, enemy air defense suppression, airlift, air refueling, and reconnaissance missions. 97