DOW-UAP-D151: Cognitive Limits on Simultaneous Control of Multiple Unmanned Spacecraft - How Many Craft Can One Operator Handle?

Source file: DOW-UAP-D151_AAWSAP-DIRD-Cognitive-Limits-on-Simultaneous-Control-of-Multiple-Unmanned-Spacecraft-December-15-2010.pdf Originating agency: Defense Intelligence Agency (DIA), Defense Warning Office; produced under the AAWSA Program Document type: Defense Intelligence Reference Document (DIRD), "Defense Futures" series; control number DIA-08-1101-001 Date: 15 December 2010 (information cutoff date, ICOD: 8 September 2010) Classification: UNCLASSIFIED//FOR OFFICIAL USE ONLY (the FOUO caveat is struck through in the banners; released in 2026) Page count: 31 VIRIN: 260918-D-D0360-1140 PURSUE Release: 6

Summary

A 31-page reference document that asks a question nobody had yet studied directly: how many unmanned spacecraft can one human operator control at the same time? It was prepared by the DIA's Technology Warning Division (DWO-4), carries the control number DIA-08-1101-001, is dated 15 December 2010, and has an information cutoff date of 8 September 2010. The author appears only as "AAP Person 73," the same pseudonym credited with DOW-UAP-D150, the DIRD on quantum and DNA computing that the same division issued five days earlier. Questions were to be addressed to "AAP Person 1," the AAWSA Program Manager. Page 2 of the PDF is blank.

Having found, in its words, no peer-reviewed article at all on remote piloting of multiple spacecraft, the paper borrows from two analogues: air traffic control (ATC) and the supervision of several unmanned air and ground vehicles. In six short chapters, two tables and six figures it reviews how mental workload is measured and what ATC and drone studies show, and sums up in three numbers: 16 craft for simple destination selection, 7 for moderately complex piloting or mission tasks, and 4 for complex heterogeneous craft. It also states that physiological signals can flag overload in real time.

The paper does not mention UFOs or UAP. The word "anomalies" appears once, as air traffic control jargon for flight errors.

Research Article

A fleet with a single pilot

The Summary (p. 5) sketches the scenario: "Space exploration 40 years into the future may include manned missions to parts of the outer solar system," perhaps with a small fleet: a trailing spacecraft of nuclear-powered electromagnets to shield the crewed part of the fleet from solar radiation, "halo spacecraft with powerful radars to scout for incoming objects," and exploration and mining craft, often out of visual range of one another. Operating such a fleet "could be economically accomplished if only one remote pilot on station at a time was necessary." The ancillary craft would be largely automated but, hundreds of miles apart, would need monitoring so that they do not simply "disappear one day during the mission like a Martian probe" (p. 6).

The central concept is "the big picture": the operator's mental representation of each craft's identity, position, mission and direction, also called situational awareness. The primary research question is whether there is a cognitive limit to the number of moving objects that can be held in it. The secondary questions are whether complexity sets that limit, and whether a real-time objective measure can show a pilot approaching or exceeding maximum capacity (p. 6). The author fences off the scope with a dry aside: "Cyborg-enhanced astrobots are a topic for another tome."

In terms of the AAWSAP Statement of Objectives (DOW-UAP-D110), the paper plausibly belongs to the "human interface" and "control" areas, with a link to "human effects" through physiological monitoring. The "incoming objects" the halo radars would scout for are not specified further, and nothing in the paper connects them to UAP.

How workload is measured

Figure 1 (p. 7) sets out a one-dimensional model the whole paper returns to: as task demand rises, an operator moves from disengagement (D), through effort to stay vigilant (A1), an optimal zone (A2) and extra effort (A3), to degradation (B) and overload (C). Chapter 2 (pp. 8-12) surveys three families of measures. Subjective: the NASA Task Load Index (NASA-TLX), with six subscales (mental, physical and temporal demand, own performance, effort and frustration), and the SWAT technique, which rates three factors on a 0-100 scale. Performance: reaction time, accuracy and secondary tasks, plus "utilization," the share of time the operator is busy; according to the paper, "at around 70% utilization performance begins to degrade" (p. 8). Physiological: heart inter-beat interval and heart-rate variability, EEG bands from delta (0-4 Hz) to gamma (30-100 Hz), the P300 potential, eye fixations and blinks, skin conductance, and stress hormones, which remain impractical for real-time use because even fast salivary cortisol assays take about 15 minutes (p. 12).

Air traffic control: the first analogue

Chapter 3 (pp. 13-22) is the core of the paper. It identifies terminal radar approach control (TRACON), where controllers vector arriving aircraft within about a 50-mile radius of the airport, as "the most cognitively demanding function in this chain" (p. 13). A footnote recalls the 1981 controllers' strike, in which 11,345 controllers were fired, and says it left the 1980s with little research on controllers.

The studies reviewed yield specific numbers. Lamoureaux (1999) predicted controllers' perceived workload 74% of the time using 81 complexity classes for pairs of aircraft (Table 1, p. 16). A stress study at Fayetteville, Arkansas, Roswell, New Mexico, and Oklahoma City found urinary epinephrine a better stress marker than heart rate. In Peiris (2005), human EEG experts identified only 6 of 101 attention lapses. In the Brookings (1996) simulation, Air Force controllers handled 6, 12 and 18 aircraft in 15-minute blocks, then 15 aircraft in 5 minutes. Figure 4 (p. 18) shows complexity hurting performance more than volume, with only the low-complexity 12-plane scenario near 100%. A 2003 neural-network reanalysis by Wilson classified the overload condition correctly "in more than 98% of the cases." Collet (2009) monitored 25 working controllers at Saint Exupery International Airport in Lyon; Table 2 (p. 19) shows the number of aircraft correlating with the NASA-TLX score (0.98), skin conductance (0.93), skin blood flow (-0.97) and instantaneous heart rate (0.98).

Two findings shape the conclusions. Wickens, in a study of 12 student pilots, found that when a conflict alarm was right more than about 80% of the time, pilots stopped verifying it, and concluded that a 20-25% false-alarm rate was optimal when the pilot is meant to work alongside the automation rather than rely on it (p. 22). A footnote compares this to setting the squelch on a CB radio. And the modeling literature (Hendy, Averty, Loft) points to decision time and complexity, not the raw number of aircraft, as the drivers of workload. The paper sets aside group models, because a lone pilot in space has "no room full of colleagues" to hand traffic to (p. 20).

Drones and the three numbers

Chapter 4 (pp. 23-26) turns to unmanned vehicles. In 2005 each U.S. Army Hunter or Shadow drone needed a team of two operators. A Dixon simulation tested whether automation could take a pilot from one-half to two aircraft, and found that autopilot and auto-alert aids helped greatly, though target-of-opportunity detection still fell from 92% to 79% with two aircraft, which the researchers blamed on four screens per vehicle (pp. 23-24). A study by Ruff concluded "that the absolute maximum number of UAVs a person could control is four" (p. 25).

Cummings, studying the retargeting of in-flight cruise missiles, measured utilization instead of complexity. The paper renders her result through the analogy of a chess grandmaster playing simultaneous games: "how many simultaneous games can the grandmaster play before he is forced to revert to cold position analysis with every new presentation of a game? The conclusion is 16" (p. 25). When Cummings added aircraft heterogeneity and queuing-theory "wait times," the maximum fell to seven (pp. 25-26).

Conclusions, forecasts and weak points

The Discussion (p. 27) ties the figure of four to working memory: it is "consistent with standard estimates of human working memory being able to handle three to five disparate objects at a time." The effective aids, whether the controllers' handwritten blocks, the stored instructions in Dixon's study or Cummings's dual displays, act as external working memory. The forecasts are modest: spacecraft-specific simulator studies within 5 years, an understanding of multitasking within 40 years, and near-perfect physiological classification of overload "within five years of specific studies being commenced" (p. 28).

The weaknesses deserve a plain statement. There is no spacecraft study at all; every number is borrowed from aircraft, missiles and drones. The mapping of the three numbers is looser than the Conclusion suggests: the 16 comes from missiles with minimal interaction between them, the 7 from Cummings's study of heterogeneous aircraft, and the 4 from Ruff's "far more demanding piloting task," yet the Conclusion attaches the 4, not the 7, to "complex heterogeneous craft." The Ruff study has no entry in the 52-item reference list, and several studies rest on small samples (8 controllers in Brookings, 12 student pilots in Wickens). Roswell, New Mexico appears only as a low-traffic ATC facility in a 1978 stress study (per the reference list), with no connection to the 1947 incident. The paper does not mention UFOs or UAP.

Significance

This DIRD shows AAWSAP's "human interface" thread in its most practical form: a literature review of air traffic control and drone ergonomics, projected onto a hypothetical crewed fleet in the outer solar system. The 16, 7 and 4 figures, and the emphasis on the "big picture," automation bias and physiological monitoring for overload, are reasonable summaries of the 2010 human-factors literature, and today they read as relevant to drone swarms as to spacecraft. The paper contributes nothing directly to the UAP question; its value lies in documenting the breadth of topics written under the program.

Key People

Locations

Key Concepts

Notable Quotes

"Piloting these multiple craft could be economically accomplished if only one remote pilot on station at a time was necessary." -- page 5

"It is not a surprise that there is little work in this specific area - in fact there were zero peer-reviewed articles in the major journals concerning remote piloting of multiple spacecraft (published in the last 30 years)." -- page 6

"For the current treatise we will consider only traditional humans as pilots. Cyborg-enhanced astrobots are a topic for another tome." -- page 6

"The question at hand is, how many simultaneous games can the grandmaster play before he is forced to revert to cold position analysis with every new presentation of a game? The conclusion is 16, and it agrees with previous work on free flight ATC." -- page 25

"The number of four is consistent with standard estimates of human working memory being able to handle three to five disparate objects at a time." -- page 27

"It was also shown in the ATC and piloting tasks that alerts need to contain a level of noise (false alarms) of 20-25% to avoid automation bias." -- page 27

"Apollo spacecraft required dozens of ground operators to monitor for system failures, and just a few years ago it required two soldiers to operate a simple reconnaissance drone (most of them still do)." -- page 27

"We expect this classification will be achieved with near perfect accuracy within five years of specific studies being commenced." -- page 28