Multi-UAS Control by a Single Operator: The paradigm shift from aircraft count to optimizing operator workload

Aircraft count has little meaning without understanding the work

“How many Unmanned Aircraft Systems (UAS) can a single operator supervise?” is a common question for m:N UAS operations. In the past, a team of operators was required to control a single UAS. With advances in automation and even the application of AI in some cases, there has been a paradigm shift. First starting with 1:1 operations where one operator could control one UAS, to 1:N where one operator controlled a few (typically 2-4 unmanned assets), to what we see today with package delivery and soon with advanced air mobility (AAM) where one or two operators supervise multiple (in some test cases up to 100) UAS. This paradigm is referred to as m:N, where m is the number of operators and N is the number of UAS. 

This evolution changes the argument for programs that traditionally have wanted straightforward numbers for system architecture, staffing plans (including personnel type and amount), cost estimates, and safety arguments. In reality, given the capabilities of today’s UAS to operate autonomously, aircraft count has little meaning without a deeper understanding of the autonomy level, operating conditions, UAS capabilities, and anticipated contingencies.

An operator overseeing a dozen aircraft on predictable, highly automated missions performs a fundamentally different job from one managing a small fleet requiring frequent replanning, communication, and intervention. Aircraft count contributes to demand, but the operator workload generated by each of those aircraft ultimately determines how many are manageable for one operator.

Identifying a realistic operator-to-UAS ratio requires understanding how personnel and automation will coordinate, what work reaches the operator, how frequently it occurs, how difficult challenges are to resolve when they occur, the cost of task switching, and what happens when several competing demands require attention at the same time.

Autonomy changes the operator’s job

UAS operations have been quickly, yet at times, quietly evolving. Today, automation handles an increasing amount of navigation, control, and routine mission functions. The human role in many cases has shifted to that of supervisory control. Depending on the concept of operations, the operator may monitor mission progress, approve or revise plans, coordinate with other people or systems, and manage abnormal conditions. Under autonomy, direct control becomes less frequent, but human interventions become more consequential because the operator must rapidly diagnose and resolve potentially serious problems as they arise. Attention, supervision, and intervention are key factors driving the capacity of the operator at any given moment.

Early research described capacity using two useful concepts: how long a vehicle can operate without attention, and how long the operator needs to interact when attention is required. Both depend on the autonomy and the operator interface, which means they need to be designed as a system rather than evaluated separately. A vehicle that rarely needs help may still consume significant operator capacity if every intervention requires a long diagnosis or difficult recovery steps.

This changes how multi-UAS capacity should be assessed. Yes, counting aircraft is easy; estimating the stream of attention-demanding events, decisions, and communications is difficult, but much more informative.

Capacity erodes when work begins to queue

Supervisory control is inherently event-driven: a vehicle requires troubleshooting, a route needs updating, a sensor task reaches a decision point, or a communication is received. The operator handles the highest priority task while other demands are delayed and relegated to a queue. The resulting queue provides a useful way to think about capacity:

• Task arrival rate describes how often work reaches the operator

• Service time describes how long the work takes

• Switching and reorientation add delay as attention moves between vehicles, displays, and mission contexts

When work arrives faster than it can be handled, response times lengthen and unattended tasks accumulate. 

Research illustrates how these variables can impact capacity. Donmez et al. (2010) held vehicle count at five and found that changing fleet composition altered how operators distributed their attention. Busier operators also took longer to switch between vehicles. Fleet composition and switching demands can therefore impact capacity without changing the number of vehicles.

Saephan and Sadler (2025) examined another queue driver: required radio communications. Their model showed how communications can produce longer waits and growing task queues as fleet size increases. The practical implication is that one recurring part of the operator’s job can become a capacity constraint even when vehicle control is highly automated.

The operator interface is a key workload driver during contingency management

Even with advances in autonomy technologies and capabilities, when automation reaches a condition it cannot resolve, the operator needs to step in. During that time, operators need more than an alert to maintain situational awareness (SA). The system must help identify which aircraft needs attention, indicate urgency, and distinguish the event from other developing issues and operator tasks. Poor alerting can create unnecessary demand that degrades vigilance or directs attention away from more urgent issues.

When an issue or emergency requires intervention, automation has been continuously engaged with the aircraft but the operator was likely monitoring the fleet at a higher level or focused on other tasks. An operator needs to rapidly orient to the issue and diagnose the right intervention. What was the vehicle doing? What changed or what outside factors were introduced? What has automation already attempted? What constraints may limit the options available? The system design challenge is to present the state, rationale, uncertainty, and likely consequences needed for the decision without needlessly spiking operator workload with unnecessary information or noisy data.

Focused attention on one aircraft also reduces the attention available for the rest of the fleet. The interface and automation need to support both levels of work: resolving the immediate problem and maintaining enough fleet-level awareness to detect the next one. This is especially important when contingencies overlap or the initial response takes longer than expected.

A related exploratory NASA cognitive task analysis conducted with the help of HF Designworks found that aviation SMEs shifted from tracking individual assets in a 12-UAS scenario toward reactive exception management in a 100-UAS scenario. Participants also relied more heavily on mission-timeline visualizations and telemetry panels (Scheff, 2021). The finding reinforces the need to design the interface around how operators locate, understand, and respond to exceptions while preserving a view of the wider operation.

Capacity will also rely on escalation and handoffs

Some contingencies may ultimately require the assistance of another operator or a role with different expertise. In time-critical situations, operators rarely have time to manually conduct a full briefing of the issue. The system must recognize when assistance is needed, transfer responsibility clearly, and provide the receiving person with enough context to contribute or assume the task quickly.

NASA research on tactical handoffs provides one example. Chandarana et al. (2022) found that operators handed off vehicles more often when more contingencies occurred and when an assisted handoff tool was available. The study does not prescribe a staffing model, but it shows why handoff design belongs in the capacity discussion: team structure and transfer support can change how an operation absorbs overlapping demand.

FAA Part 108 looks beyond nominal aircraft count

The FAA’s 2025 Part 108 Notice of Proposed Rulemaking (NPRM) formalizes this paradigm shift to supervisory control and reinforces why capacity and workload become major focal points for management of unmanned fleets. In Part 108, oversight shifts from an active "remote pilot" role manually handling assets to an Operations Supervisor or Flight Coordinator responsible for monitoring automation health, strategic deconfliction, and exception handling. Proposed §108.210 would allow ratios greater than 1:1 through a method “acceptable to the FAA”, while limiting a flight coordinator to the number of aircraft they can handle during “normal, abnormal, and emergency conditions.” The FAA also explains that a manufacturer’s maximum ratio cannot fully account for an operator’s procedures, operating conditions, and human factors. 

You can read more about Part 108 and what it says about workload in our blog post, How Workload Fits Into the FAA's Proposed Part 108 Ruleset.

The FAA is not the only organization developing UAS standards. ASTM will soon finalize standards for multi-UAS ground control station design. SAE maintains standards for multi-unmanned systems control-segment architecture. DronesQuad, an effort involving airworthiness experts from FAA, EASA, TCCA, and ANAC, has published high-level product-safety criteria intended to guide standards development.

Treat the ratio as a claim to validate

Because capacity is driven by the mission, automation, interface, and human role, any proposed operator-to-UAS ratio should first be treated as a claim to validate. A reliable evaluation starts by defining the roles of human and autonomy and required outcomes. What decisions remain with the operator vs. with autonomy? Which events and contingencies require intervention? How quickly must the system respond, and what mission or safety performance must be preserved?

Next, characterize the work. Task analysis and workload modeling can estimate event frequency, interaction duration, switching, communications, and plausible overlap. Modeling is valuable for comparing concepts before every element exists, but its assumptions should be calibrated with human data where possible.

Human-in-the-loop evaluation can then test representative nominal and off-nominal conditions. Vehicle count is one experimental variable among many. Mission performance, delayed or missed events, response time, error, workload, situation awareness, attention allocation, and reliance on automation provide a more complete view of how well the overall concept of operations is performing. Testing should become more realistic as the design matures and should include the combinations of events most likely to expose an operator capacity limit.

Any resulting ratio is tied to the conditions: the mission, human role, automation configuration, interface, communications, environment, training assumptions, scenarios tested, and acceptance criteria. Without that context, the number is difficult to reproduce and easy to misuse.

HF Designworks supports human-performance modeling, human systems integration, interface design, and human-in-the-loop evaluation for complex aviation and defense systems. If your team is working on solving multi-UAS operations, contact us for more information.

Sources and Related Reading

Bhaskara, A., et al. (2021). Effect of Automation Transparency in the Management of Multiple Unmanned Vehicles (https://doi.org/10.1016/j.apergo.2020.103243). Applied Ergonomics, 90, 103243.

Chandarana, M., et al. (2022). Streamlining Tactical Operator Handoffs During Multi-Vehicle Applications (https://doi.org/10.1016/j.ifacol.2022.10.235). IFAC-PapersOnLine, 55(29), 79–84.

Crandall, J. W., Goodrich, M. A., Olsen, D. R., Jr., & Nielsen, C. W. (2005). Validating Human–Robot Interaction Schemes in Multitasking Environments (https://doi.org/10.1109/TSMCA.2005.850587). IEEE Transactions on Systems, Man, and Cybernetics - Part A, 35(4), 438–449.

Cummings, M. L., & Guerlain, S. (2007). Developing Operator Capacity Estimates for Supervisory Control of Autonomous Vehicles (https://pubmed.ncbi.nlm.nih.gov/17315838/). Human Factors, 49(1), 1–15.

Donmez, B., Nehme, C., & Cummings, M. L. (2010). Modeling Workload Impact in Multiple Unmanned Vehicle Supervisory Control (https://doi.org/10.1109/TSMCA.2010.2046731). IEEE Transactions on Systems, Man, and Cybernetics - Part A, 40(6), 1180–1190.

Mercado, J. E., et al. (2016). Intelligent Agent Transparency in Human–Agent Teaming for Multi-UxV Management (https://doi.org/10.1177/0018720815621206). Human Factors, 58(3), 401–415.

Monk, K. J., Rorie, R. C., Brandt, S. L., Sadler, G. G., & Roberts, Z. S. (2019). A Detect and Avoid System in the Context of Multiple-Unmanned Aircraft Systems Operations (https://ntrs.nasa.gov/citations/20190027572). NASA/AIAA.

Saephan, M., & Sadler, G. G. (2025). A Queuing Theory Approach to Pilot-Controller Coordination for m:N Operations (https://ntrs.nasa.gov/citations/20240015006). NASA/AIAA.

Scheff, S., et al. (2026). Use Cases, Emerging Technologies, and Beyond Visual Line of Sight: Needs, Gaps, and Understanding Processes for sUAS (url) NASA Technical Memorandum (pending publication).

Scheff, S. (2021). HAT m:N Cognitive Task Analysis (CTA) (https://ntrs.nasa.gov/citations/20210011503). NASA Multi-Vehicle Control (m:N) Working Group presentation.

- Wohleber, R. W., et al. (2019). Vigilance and Automation Dependence in Operation of Multiple Unmanned Aerial Systems (UAS): A Simulation Study (https://doi.org/10.1177/0018720818799468). Human Factors, 61(3), 486–505.