Spectral Autonomy

A-GRA: Interfaces for mission autonomy

Part of the Open Architecture Field Guide.

The Autonomy Government Reference Architecture (A-GRA) defines how mission-autonomy software exchanges information with a vehicle and its surrounding systems. It gives autonomy software a common integration boundary across implementations.

In this guide

Mission decisions need information from the platform

Mission-autonomy software uses a plan and an estimate of the current situation to decide what a platform should do next. To make those decisions, it needs information about the vehicle, available mission functions, and other participating systems. It also needs a defined way to communicate its decisions to the components responsible for carrying them out.

The Autonomy Government Reference Architecture (A-GRA) specifies the interfaces around mission autonomy. This makes the decision software a defined participant in a larger system. The autonomy developer works against the information and interactions at that boundary, while the platform integrator connects those interactions to the actual vehicle and equipment. [1]

The distinction is useful because mission decisions and vehicle execution are different responsibilities. Mission autonomy can request behavior based on its objectives and current information. The vehicle-side functions retain their own responsibilities for executing and validating requests. An interface describes how the two sides coordinate without placing all vehicle implementation details inside the mission decision logic.

Six interfaces around mission autonomy

The Autonomy Starter Kit 5.0a describes six Level 1 interfaces. “Level 1” refers to the external system boundary around mission autonomy. Each interface organizes a different relationship with the surrounding system. [1]

The vehicle interface exchanges vehicle state, subsystem status, and requests with the functions responsible for the platform. The mission-systems interface connects autonomy with onboard capabilities and their reports. Together, they provide information about what the platform is doing and which mission resources are available.

The command-and-control interface carries direction from the systems or people assigned those responsibilities. The peer-to-peer interface supports exchanges with other participating platforms. These interfaces place the autonomy package in a wider coordination structure instead of treating it as an isolated decision-maker.

The mission-planning interface supplies plans and configuration needed for execution. The mission-debrief interface supports recording and reviewing what occurred. These two relationships connect preparation, execution, and subsequent analysis through defined information exchanges.

A-GRA mission-autonomy boundaryMission autonomy exchanges information with vehicle, mission systems, peer, command-and-control, plan, and debrief interfaces.Mission autonomyVehicleMission systemsCommand and controlPeers · plan · debrief

Scroll horizontally to inspect the diagram.

Figure 1. The ASK 5.0a interface volumes define the exchanges around mission autonomy. Connecting lines show interface relationships, without implying an execution sequence.

Following a change in available information

Consider an illustrative situation in which an onboard capability reports that its availability has changed. The mission-systems interface provides that information to mission autonomy. The autonomy software can incorporate it into its situation estimate and reconsider the choices available under the current plan.

If the revised decision requires a change in platform behavior, the corresponding request passes through the vehicle interface. The vehicle side evaluates and responds to the request according to its responsibilities. The distinction between requesting behavior and executing it keeps the interface meaningful: both sides exchange defined information while retaining their own implementation and decision obligations.

The example shows why the architecture specifies interactions as well as individual messages. A request has to be related to the status and response that follow it. The Starter Kit uses messages from the Universal Command and Control Interface (UCI), with mission-autonomy extensions, to describe these exchanges. Its compliance approach examines the required messages and sequences. [1]

Developing autonomy and platform integration separately

With the boundary defined, an autonomy team can evaluate its logic against implementations of the required interfaces. A platform team can develop the corresponding integration without having to incorporate that logic into every equipment-specific component. Later changes can be located on the side of the boundary whose responsibility has changed.

For example, improving how autonomy interprets reports may preserve the interfaces through which those reports arrive. Replacing an onboard capability may instead require work on the provider side while preserving the information supplied to autonomy. The architecture makes the relationship between those changes explicit.

This is where A-GRA meets AMS-GRA. A-GRA describes exchanges around mission autonomy; AMS-GRA describes mission-system functions and their integration boundaries. A service can be provided through one architecture and used by the decision software described by the other. Their scopes explain how the requesting and providing responsibilities meet.

References

  1. U.S. Air Force, Autonomy Starter Kit 5.0a: Start Here Guide and interface volumes, April 2026. Distribution A.