by U.S. Army Dr. Elizabeth Mezzacappa, Acquisition Support Center
October 3, 2018
The use of increasingly sophisticated tools over time is one of the defining characteristics of humankind. This trend’s potential has been imagined in literature, often through fictional inventors such as Marvel Comics’ Shuri (from “Black Panther,” 2005) and Tony Stark (“The Invincible Iron Man,” 1963), Ian Fleming’s Q (James Bond series, 1958) and even Isaac Asimov’s Susan Calvin (“I, Robot” series, 1945). These illustrate humans teamed with devices that are more than merely tools, but are engineered “entities”—robots, drones, swarms and other tools that their human creators have endowed with very human capabilities. Which means that humans will adapt quickly to these tools—and the better they are designed, the more quickly humans will adopt and evolve them.
From cybernetic enhancements to artificial intelligences based on human neuroscience, these technological developments require a merging of engineering and psychology. It is one thing to create the tools we need today. It is entirely another to envision the future and create the tools we will need then.
We know the future means humans even more closely teamed with their tools. A significant question for the engineers, scientists and psychologists who are developing future tools is, which fantastical elements invented by Q, Stark, Calvin or Shuri can (and should) be engineered in our real universe, with our real physics, and real flesh-and-blood Soldiers?
Combat occurs increasingly in complex urban centers and among noncombatant populations, so engineers must develop armaments and protection optimized for these settings. Most difficult is “engineering” the human psyche and human flesh into these created machines and integrating so seamlessly that the technological “magic” becomes mundane, as easy as putting on pants.
Engineers tend to lack a background in people sciences like psychology, so how do engineers generate the data about humans needed to create systems of human-machine symbionts ready for war in unfamiliar territories? Tactical behavior research is one way, an approach that looks to understand and improve human and machine performance in tactical, combat situations through close collaboration between human behavioral scientists and materiel developers who build armaments and other tools for Soldiers. Since 2004, the Tactical Behavior Research Laboratory (TBRL), of the U.S. Army Research, Development and Engineering Command (RDECOM) Armament Research, Development and Engineering Center (ARDEC), has conducted human tactical behavior research.
The laboratory’s research looks at humans at both ends of the barrel. Focus areas include:
HIGH-FIDELITY LABS AND TEST BEDS
For a Soldier standing watch, the minivan barreling down on the checkpoint is a life-or-death situation in which the Soldier has seconds to decide whether to open fire. Soldiers in a convoy of Army trucks facing a crowd of angry townspeople—blocking a road, chanting and throwing rocks—have to disperse the gathering of civilian men, women, elders and children to complete the mission.
The human science—the psychology of the Soldiers in these scenarios—is critical to understand. This can’t be done by conducting research in the typical one-room psychology laboratory of a university or research institute testing undergraduate student subjects, or even the researcher’s own co-workers.
To understand that psychology, we need research conditions that mimic these settings. That’s what distinguishes tactical behavior research from typical psychology experiments. Behavioral science theories are used to guide the engineering of the test beds to create the appropriate psychological (perceptual, motivational, social) conditions for the experiment and to capture the appropriate data.
TBRL has created a number of indoor and outdoor laboratory conditions that simulate real-world tactical scenarios at its facility at Picatinny Arsenal, New Jersey. A sample of the test beds includes:
The largest test bed, the Squad Performance Test Bed, consists of both a large outdoor area of 700 by 500 meters (about eight football fields) and an indoor test bed. The outdoor area has instruments to capture behaviors of fire teams, squads, platoons or other groups during outdoor warfighter battle drills. Instrumentation includes video cameras and motion-capture sensors to record Soldier responses in a react-to-contact battle drill. The indoor structures were custom-built for room-entry testing and are modifiable to be center- or side-fed rooms, since door location determines where the Soldier points the gun.
REAL SCIENCE, VIRTUAL TEST BEDS
One might ask: If test conditions are supposed to be close to combat conditions, how can TBRL test for urban or subterranean settings that don’t exist near rural New Jersey? New levels of both realism and experimental control are now achievable with immersive virtual-, mixed- and augmented-reality laboratories. TBRL’s first virtual-reality laboratory was built in 2010. Now in its third iteration, the testing facilities include multiple 360-degree mixed and augmented virtual-reality simulators in 30-by-30-foot octagons. In addition, TBRL has integrated a virtual-reality headset system that fully immerses viewers into the scenario. The virtual environments are developed in-house by ARDEC’s Gaming and Interactive Technology and Media group, which allows researchers access to all aspects of the system to modify and extend their capabilities for experimentation.
One extended capability is achieved through combining the virtual environment methods with motion capture abilities—avatars. That is, the test bed virtual environment is brought to life by incorporating a wide range of avatar behaviors within the computerized scenery. Computer-generated characters and avatars greatly extend the capabilities for social-psychological experimentation into human-human or human-entity teaming. In the golden age of group dynamics studies, “stooges” (i.e., research actors with scripts) were used, unbeknown to subjects. These stooges acted to create a controlled social situation (think of the famous Milgram conformity experiment, in which the stooge was instructed to yell in pain when subjects turned up a knob that looked like it was delivering electrical shock, in order to test the subjects’ obedience to authority figures).
In place of stooges, characters and avatars can be programmed to behave in any manner and take on any appearance that artists can render in programming. With avatars, we can, for example, conduct a Soldier-robot interaction experiment without the time and expense of building a real functioning robot. Artists could render a humanoid-looking metal entity that moves and speaks through a researcher’s movements and speech. In the virtual environment, then, Soldiers are led to act as if they were interacting with robots. Researchers could learn about how best to build robots so that Soldiers will work with them, through observing these virtual interactions and providing data to inform robot design requirements.
In a similar way, experimentation with weapons that do not exist is made possible by research in a simulator. Through software, programmers can create future weapons in the simulators, then operators use these devices within combat scenarios. For example, testing might examine the effect of increasing weapon range versus area covered for use in an urban environment, where distances are more limited than in open fields. Virtual experimentation with simulated weapons allows designers to gather lethality and other performance data and feedback from operators before bending metal. Human experimentation in the virtual environment allows materiel developers to verify and validate novel concepts of armaments such as swarms of drones or directed energy weapons, as well as to identify performance requirements, especially lethality, in advance. In this way, researchers can chart the progress toward future weapons with more certainty.
A RADICAL DEPARTURE
A reading of any behavioral science journal article reveals quickly that typical psychological research is simply not configured to answer engineering questions. Research psychologists strive to reveal universal precepts of human behavior. Materiel developers yearn for characterization of a specific device. Tactical behavior experimentation must bend psychological science in service to engineering, and adopt the mindset and constructs of acquisition science, such as metrics of lethality, verification and validation, cost and capability trade-offs, analyses of alternatives, benchmarking and comparative testing of specific devices.
For example, one experiment planned by TBRL will research the relationships among Soldier cognitive fatigue, number of drones controlled and number of targets destroyed over a simulated mission to develop an algorithm that explains the connections among those variables. With this type of data, analysts can conduct trade-off analyses—for example, balancing the cost of building the optimal number of drones and Soldier-drone interfaces versus the lethality of the drone swarm versus the cognitive demands and stress on the operator.
Current regulations require that human-factors professionals—who primarily assess designs for ergonomic flaws and related Soldier performance concerns—be consulted at all developmental research and operational testing phases of the acquisition cycle. Based on our experience at TBRL, we propose a more radical solution—that each of the research engineering centers establish its own dedicated behavioral scientist laboratory to conduct the relevant engineering-focused human science experimentation. This early collaboration would then complement the independent, third-party role of evaluation later provided by the human-factors specialist. By “embedding” behavioral science laboratories in all Army research, development and engineering centers, the right human research is done effectively to support the development of equipment for Soldiers. However, a required precursor for engineering-focused human research is perhaps the most challenging aspect—joint experimentation efforts between engineers and psychologists.
In the last few years, TBRL’s capabilities have come to the attention of the larger RDECOM ARDEC engineering community. Armament engineers have questions that require behavioral science methods and research designs and analysis. Behavioral scientists begin by working with materiel developers on articulating the knowledge gap, then translating the knowledge gap into a behavioral science research question. More discussions follow, resulting in designing the experiment and analyses to generate human data that is needed to answer the question, describe the requirement or guide design. Engineers assist in the actual running of the study as well.
Materiel developers now come to the laboratory with questions that can be answered only through human-subject research, which requires experimentation that is approved and overseen by boards that ensure ethical conduct of research. Behavioral scientists are well-trained in the principles of the ethical conduct of human-subjects research, a topic possibly quite foreign to engineers. Therefore, in preparation for running human experimentation, engineers also take the required human-research ethics training.
There are many benefits to collaborations. Joint research between engineers and psychologists aligns with the cross-functional teaming principles outlined in the Army’s modernization priorities. Moreover, the close collaboration of engineers and psychologists in tactical behavior research addresses transition problems identified in the 2015 publication “Soldier Squad Performance Optimization.” This report cited the challenges of bringing behavioral science data to customers—both to Soldiers and to engineers who build Soldier equipment. At least the second challenge can be resolved when materiel developers pose the engineering research questions and work with behavioral science on the experimentation to answer them. Joint research also mitigates the risk that promising technologies won’t make it to Soldiers, a risk cited by the Army Science Board 2017 study “Improving Transition of Laboratory Programs into Warfighting Capabilities Through Experimentation.”
How engineers and psychologists engage in joint research is demonstrated in the current laboratory efforts in the Armament Virtual Collaboratory Environment project. The work is a collaboration with the ARDEC Operational Analysis Branch to collect human performance and psychophysiological data in the virtual environment to support development of artificial intelligence that could aid the dismounted Soldier. That is, the experiment gathers detailed information on how someone is doing while performing a task, not only physically but also psychologically. In turn, those data are submitted to machine-learning analysis to inform the development of devices that are trained to “think” the way the gunner thinks.
Specifically, engineers approached the Gaming and Interactive Technology and Media group and TBRL to design and demonstrate a behavioral experiment to identify characteristics of potential targets that lead to Soldiers’ decisions in the battle. The intended long-term outcome of the work will be a Soldier-armament interface with advanced fire controls, including optics and displays that enhance system lethality. This is the research we need to get to Jarvis, Tony Stark’s machine assistant, and the target acquisition schematic he projects onto Iron Man’s visor.
These data and other results must be gathered to answer fundamental questions for future warfare. What new structures of command and control must be configured between human and engineered entities? How should the labor be divided between them? Do we enhance the human brain and muscle or juice up the hard drive and armature? Or: Who (or what) pulls the trigger? Only research with both human behavioral and engineering considerations can answer these questions.
The successful blending of engineered entities and human entities is achievable only through collaborative science and experimentation between engineers and psychologists. Those robots won’t build themselves. Not yet.
DR. ELIZABETH MEZZACAPPA is the human research lead at RDECOM ARDEC’s Tactical Behavior Research Laboratory, where she has worked since 2007. She also is an assistant professor at the Army’s Armament Graduate School. She holds a Ph.D. in medical psychology from the Uniformed Services University of the Health Sciences and B.A. degrees in psychology and biology from the University of Pennsylvania.