ACE HIGH LEVEL ARCHITECTURE
A Multi-Agent System (MAS) is a collection of agents that interact and work with each other where each agent is an entity that works in an organization, perceives events, and takes actions in an environment.
The MAS serves two purposes: a foundation for the Knowledge Platform (KiP) and a stand-alone product that is used outside KiP for non-AF Autonomy related work. The KiP provides an ecosystem for creating solutions.
The MAS is a small light-weight code base that is fast and flexible. It provides all of the foundational agent components that are common to all agent operations: spawning, organization, sensing, acting, feedback and communication. When used with the KiP, it provides Peer, Task, and Cognitive Flexibility. It supports both software-based solutions as well as embedded hardware-based solutions in order to meet the demands of the Air Force.
Per Autonomous Horizons: The Way Forward (Zacharias et al., 2018), autonomous systems are Artificial Intelligence-based systems that provide peer, task, and cognitive flexibility. Peer flexibility means that agents can change their relationship with other agents (human or machine) and seamlessly take on a subordinate, peer, or supervisor roles based on need. Task flexibility means that agents can change the nature of the task they are working on in order to achieve their actual (or implied) objectives. Cognitive flexibility means that agents can change their approach to solving a single (multitude of) problem(s).