Report on Human Factors Issues Likely to Affect Air-Launched Effects

This report reviews human factors research on the supervision of multiple unmanned vehicles (UVs) as it affects human integration with Air-Launched Effects (ALE).

February 1, 2024

U.S. Army Combat Capabilities Development Command Analysis Center, Fort Novosel, Alabama

Figure 1. Raytheon’s Coyote Block 3 (left) and Area-1’s Altius 700 (right) are exemplars of the ALE to be developed.

Air-Launched Effects (ALEs) are a concept for operating small, inexpensive, attritable, and highly autonomous unmanned aerial systems that can be tube launched from aircraft. Launch from ground vehicles is planned as well, although Ground-Launched Effects are not yet a requirement. ALEs are envisioned to provide “reconnaissance, surveillance, target acquisition (RSTA), and lethality with an advanced team of manned and unmanned aircraft as part of an ecosystem including Future Attack and Reconnaissance Aircraft (FARA) and ALE.” A primary purpose of ALEs is to extend “tactical and operational reach and lethality of manned assets, allowing them to remain outside of the range of enemy sensors and weapon systems while delivering kinetic and non-kinetic, lethal and non-lethal mission effects against multiple threats, as well as, providing battle damage assessment data.”

Key features of the ALE are adherence to the Modular Open Systems Architecture so that components rely on nonproprietary interfaces and the Scalable Control Interface, an application programming interface intended to allow control through any command device adhering to the protocol, including both air and ground forces.

Over the past three years, in a series of experiments and demonstrations known as Experimentation Demonstration Gateway Event (EDGE) and Project Convergence (Convergence), ALE prototypes have demonstrated increasing sophistication and capability. At Convergence-20, six ALEs covered a range of 62 km relaying RSTA information back to manned assets at standoff distances. Four were air launched while two demonstrated a new ground launch capability. In EDGE-21, Altius 600 ALE prototypes were launched as a small swarm and controlled from a Blackhawk helicopter. By EDGE-22, four waves of seven ALEs each (28 total) were ground launched, demonstrating the “flood the zone” capability envisioned by Brigadier General Rugen, head of the modernization team for Future Vertical Lift (FVL).

As the Army moves toward increasingly complex Multi-Domain Operations, the ability to effectively deploy large numbers of unmanned systems becomes crucial. This research analyzes human–autonomy interaction through the prism of scalability to better understand how unmanned systems might be deployed and supervised at scales needed to impact modern warfare. Our analysis finds three broad categories of control: autonomous coordination in which UVs are controlled as a team, quasi-independent control in which UVs are controlled largely independently of one another, and coordinated control in which operator(s) must directly coordinate UVs. We conclude that direct coordination is infeasible for realistic applications, leaving automated coordination and quasi-independent control as feasible options.

Quasi-independent control, in which operator(s) direct one or more UVs at largely independent tasks, such as area search with minimally overlapping patterns, offers many advantages in that multiple UVs can be controlled in sequence and scaled beyond a single operator’s limits, simply by adding additional operators. Depending on the arrival rates of events requiring operator attention such as targets, UVs might be controlled out of call centers, taking advantage of load balancing to increase the UV to operator ratio. This option also has the advantage of relying largely on COTS drone technology with map and first person displays for which there is already a body of research and experience requiring only protection from electronic warfare to be fielded.

Automated coordination can take varied forms, each of which present different human control issues. Simple forms of coordination, such as consensus algorithms (swarms), are fully scalable, immune to attrition, but limited in behavior. Rule or optimization-based coordination is limited in scale and may be brittle but can produce complex coordinated behaviors. A useful way of thinking about this problem is to use pre-optimized behaviors or “plays” for recurring locally coordinated behaviors, such as Battle Damage Assessment, while coordinating larger groups through scalable methods, such as consensus algorithms.

This work was performed by Jamison Hicks, DEVCOM Analysis Center; Michael Lewis, University of Pittsburgh; and Katia Sycara, Carnegie Mellon University. For more information, download the Technical Support Package (free white paper) below. DACTR-078