Frontier, a decision engine for designing stable adaptable complex systems: Adaptive framework

Frontier is a web-based framework for concurrent multidisciplinary collaboration on the analysis and design of system architectures associated with systems of systems. Supported by DARPA TTO System F6 (Future, Fast, Flexible, Fractionated, Free-Flying spacecraft architecture), the aim is to develop a framework that can be used to analyze alternative approaches to the use of space. In particular, Frontier aims to support analyses involving radically different viewpoints and approaches that are nonetheless equally legitimate views of the system or architecture under study. Furthermore, Frontier aims to federate these different approaches in a single extensible framework of interoperability standards. Frontier provides a methodology for integrating resources such as engineering tools and libraries for use on Frontier. The plan is to allow Frontier users to use their existing tool chains to produce and consume resources for a world-wide web-enabled aerospace market. Eventually, the goal is to enable aerospace stakeholders, analysts, suppliers, and others to work on distributed projects ranging through entire life cycles, from preliminary studies, collaborative bidding, design, implementation, operation, maintenance, decommissioning, replacement, and so forth. Frontier makes aggressive use of emerging standards and web technologies, especially the semantic web, and with its focus on open source methods it aims to become the way to exchange and collaborate on aerospace data and projects. A semantic bus is used to discipline the exchange of information and to manage translations between different contexts. Social media are available providing multiple ways for Frontier users to interact with each other and the tools and resources available on Frontier. Being a web-based and standards-based framework, new technologies can be brought into Frontier enabling new modalities of use and interaction. A central element of Frontier is a synthetic neural system (SNS) that monitors the performance of system tools and aids in both the selection of tools and the development of resources for use on Frontier. Since a key use case of Frontier is the development of analyses to decide between candidate architectures, the SNS aids in the selection and development of analysis tools and even the elements of the candidate architectures themselves. The behavior of the SNS is implemented with Neural Basis Functions that interact via an Evolvable Neural Interface. Chief among these NBFs is an executive function, a ToolUser, and a social interface to the user-facing human interface. Linked together and governed by the Stability Algorithm for Neural Entities (SANE), these NBFs provide the core adaptability of the system. Critical for the future development of space is creation of an adaptable and flexible system into which multiple stakeholders can provide modular capabilities that address their individual needs in as coherent and stable a manner as possible. A stable process for determining candidate architectures that meet as many goals of a community or marketplace of space services is a central goal of the DARPA F6 project. The pathway towards this goal is the SANE approach developed at NASA. In this work, Frontier researchers describe how the development of flexible, viable systems of aerospace systems and services is enabled by the use of artificial intelligence technology originally developed to address the needs of autonomous spacecraft operation in uncertain and irregular (chaotic) situations.