GA-ASI Demonstrates MQ-20 Avenger's Autonomous Team Coordination
GA-ASI and the U.S. Navy demonstrated collaborative autonomous mission planning using the MQ-20 Avenger, showing how autonomy can interpret commander intent and adapt plans in real time with mixed human-uncrewed teams.

General Atomics Aeronautical Systems, Inc. (GA-ASI) and the U.S. Navy’s Strike Planning and Execution Systems program office (PMX-281) demonstrated collaborative autonomous mission planning on September 10. The test used GA-ASI’s MQ-20 Avenger® Autonomous Collaborative Platform (ACP) at the Gray Butte Flight Operations Facility near Palmdale, California, and the Mission Operations Center in Poway, California.
The MQ-20 Avenger demonstrated its ability to understand commander intent and autonomously coordinate with human and unmanned teammates across land, air, and maritime threats. The mission autonomy software could understand a human commander’s intent and exercise autonomous judgment. It autonomously coordinated a diverse team of Group 5 uncrewed aircraft with human-crewed teammates.
The system proved it could understand a human operator's instructions and carry them out with the team. Mission autonomy software worked across the full kill chain using passive sensors, autonomy, and coordinated operations. The test focused on demonstrating that a mission designed by a human operator can be transferred to an unmanned aircraft to interpret objectives, coordinate actions, and adapt behavior.
GA-ASI’s Tactical Autonomy Ecosystem (TacACE) software received mission information and reasoned about it. TacACE executed assigned tasks and modified the plan when scenario conditions changed. In an operation, the unmanned aircraft could receive a task within a larger mission and determine how to accomplish it based on available information. If the situation changes, autonomy can adapt execution without manual re-entry of orders for every movement.
TacACE and A-GRA enable seamless human-machine teaming
GA-ASI’s TacACE software used A-GRA-compliant data to translate mission objectives into actionable plans, allowing real-time adaptation as scenarios evolved. The system extended the Navy’s planning systems with Autonomy Government Reference Architecture (A-GRA)-compliant mission context. TacACE provided platform information for integrated human-machine mission planning, command and control, autonomy configuration, simulation, and post-mission analysis.
In an exercise called Gray Flag 2026, the Navy’s baseline mission planner was augmented with A-GRA-compliant data products. The A-GRA-compliant data products exposed mission objectives, constraints, and threat context to the autonomy layer. TacACE could reason over the plan and adapt it in real time as the scenario evolved. The system supported planning and coordination of missions involving multiple aircraft and other mission participants. Operators could define objectives, tasks, priorities, and coordination rules across the force.
The Navy added information to its planning system that can be interpreted directly by autonomy software. The information included mission objectives, priorities, limitations, and threat context. A-GRA functions as a common language between human planner and autonomous system.
Live-virtual-constructive framework enabled realistic evaluation
The demonstration incorporated a Live, Virtual, and Constructive (LVC) framework. LVC environments connect physical aircraft with simulated platforms and constructive entities. LVC allowed evaluation of multi-ship coordination, autonomy behaviors, mission timelines, and operator interactions.
Participants included live, virtual, and constructive entities. The system used passive sensors, autonomy, and coordinated operations among live, virtual, and constructive participants. The LVC framework supported scalable mission rehearsal and operational evaluation. The test demonstrated coordination between multiple vehicles, including manned aircraft and simulated participants. The system was used to coordinate participants against land, air, and maritime threats. Some participants were real aircraft; others were represented through simulations. The scenario combined a real aircraft with virtual and simulated elements.
Gray Flag 2026 provided operationally relevant testbed
As the Navy’s largest-ever test exercise, Gray Flag 2026 offered a realistic setting to validate autonomy in joint, multi-domain operations. The Gray Flag demonstration is part of GA-ASI’s ongoing campaign of increasingly complex, operationally representative flight tests. The campaign combines live aircraft, mission systems, sensors, communications, autonomy software, high-fidelity digital twin models, and LVC capabilities. The campaign advances collaborative unmanned operations and informs development of mission-ready ACP capabilities.
The flight marks another milestone in warfighter adoption of ACP. It advances operational readiness and production maturity of autonomy-enabled, AMS GRA-compliant aircraft. Gray Flag 2026 was the US Navy’s main large-scale test exercise and its largest edition held to date.
| Metric | Gray Flag 2026 Detail |
|---|---|
| Sorties | More than 750 |
| Flight Hours | More than 2,000 |
| Participants | Some 3,000 |
| Organizations | Over 150 |
| International Partners | Australia, Canada, United Kingdom |
| Domains Involved | Air, naval, and submarine |
Rear Admiral Keith Hash, commander of the Naval Air Warfare Center Weapons Division, explained the purpose of finding and fixing problems during testing. Michael Roberts of GA-ASI said, "Working together, we can get the Navy to the next level."
The next step will be to progressively increase scenario complexity by adding more aircraft, sensors, threats, and participants.





