Drone swarming, as currently understood, doesn’t exist yet. True swarming requires autonomous, distributed systems that adapt in real time, not just massed robotic maneuvering, reports Emma Bates and S. Ryan Quick of War on the Rocks. While the U.S. military focuses on hardware, achieving real swarming requires a shift to distributed systems infrastructure. AI alone can’t enable swarming; it’s the system architecture that matters. To stay ahead, the military must prioritize this technology in its acquisition strategies. They write:
Despite the hype, drone swarming doesn’t exist yet. That’s because the U.S. Department of Defense has been focused on platform capability inputs like hardware, manufacturing, and GPS, while so far neglecting the architectural question of how drones are supposed to work together. The “swarming” on offer today consists of robotic maneuver en masse. However, a transformative strategic leap forward is within reach: resilient, collaborative, autonomous problem solving at machine speed, without any single point of failure. Achieving that vision of swarming will require shifting acquisition priorities toward distributed systems infrastructure, not just quantity or quality of platforms.
What Swarming Is and Isn’t
We are experts in distributed systems, collaborative autonomy, and military modernization strategy. We’ve both had security clearances. We’ve interviewed at least a hundred subject matter experts across the public and private sectors. We have seen no evidence of true drone swarming, anywhere. Those mind-blowing Chinese drone light shows? Not swarming. Leader-follower autonomous teaming? Not swarming. One hundred drones operated by a single person? Not swarming. A genuine swarm is singular, not plural. It is overwhelming not just in its scale, but in its unity and resilience. It adapts intelligently to changing circumstances at machine speed. It achieves more than the sum of its parts. It’s a distributed system.
The U.S. defense industry has so far failed to deliver distributed systems for useful, resilient, collaborative swarming behavior among autonomous platforms. […]
More broadly, to avoid fielding vulnerable and ineffective distributed military systems, the Department of Defense should clearly articulate initiatives to research, test, and procure distributed systems infrastructure independent of cloud provider dependencies. This is necessary because relying on distributed systems infrastructure in the cloud requires more bandwidth than deployed operators can count on. It also constitutes a single point of failure. Program Executive Offices and requirements developers should stipulate that distributed software infrastructure be cloud-independent and locally self-contained.
The field of research into cloud-independent, local distributed systems is shockingly underdeveloped, and the demand signal for it is vague and subject to confused messaging. But that capability is a foundational technological prerequisite to the U.S. military’s modern warfighting approaches. Indeed, each of the services’ official operating concepts are based on the assumption that some distributed infrastructure layer will enable dispersed groups of platforms to coordinate at machine speed. […]
Read more here.
