Army researchers are exploring how humans interact with an intelligent system—specifically, an exoskeleton boot that adapts to and assists its wearer in real-time—to ultimately minimize soldiers’ training time with future technology-enabled gear.
The wearable machines could also one day pose significant enhancements to military members’ performance, according to officials involved in studying how people and autonomous technologies adjust to one another when operating together.
“The assistance of a mechanical muscular-assist device will enable our soldiers to carry loads for a greater amount of time and distance. This will increase their ability to remain on patrol without need for resupply,” Army Research Laboratory Senior Enlisted Leader Sergeant Major Luke Blum said this week. “It will also decrease their time to target. This will make them more lethal on the battlefield.”
Blum and ARL Research Scientist Dr. J. Cortney Bradford briefed Nextgov on this ongoing pursuit to help advance the function and adoption of future-facing combat technologies.
Gaps Going Forward
Bradford spent years diving deep into human locomotion, or walking, using biomechanics and neuroscience approaches.
“Studying human-tech teaming and co-adaptation during locomotion allows me and my collaborators to leverage our expertise to answer these basic science questions to improve human-agent teaming,” she explained. The scientist added that basic science research inside the lab is shaped by “predictions of what the battlefield of the future will look like and what scientific gaps we need to fill to ensure operational overmatch.”
Looking ahead, Bradford noted that it’s likely soldiers will eventually be working side by side with many different instantiations of smart tech—or “technology that is capable of learning and adapting”—in warzones. Historically, military insiders have had to learn to use technological tools. But smart tech is different, Bradford indicated, because it offers the potential to adjust to each new user. Such capabilities could enable soldiers to use novel tools, like exoskeletons, without embarking on long training periods.
In order for this smart tech to adapt to its human teammates, however, it must have some understanding of what Bradford deemed the “dynamic soldier.”
“This is the critical scientific gap that we are aiming to fill with this line of research: What information from the soldier is critical to optimizing adaption between tech and the soldier?” she said. “We don’t want the soldier to have to overtly control the tech: that takes them away from other critical tasks.”
To confront this opportunity area, ARL is steering a study into individuals’ brain signals, muscle signals, movement profiles and walking performance metrics, and essentially “track their state” as they interact with the exoskeleton. The signals could offer the device a better understanding of the human to improve how it interacts supporting them, while the research also sheds light on how people get accustomed to wearable, intelligent technology.
A Spring in Soldiers’ Step
A non-Army-made product—the Dephy ExoBoot, an autonomous exoskeleton for walking augmentation—is one element at the core of this study. ARL does not have a contract with the robotic designs firm Dephy but purchased the device directly from it.
That ExoBoot, according to Bradford, “looks like a regular boot,” except there is a strut that extends from the side of it and attaches to the lower leg just below the knee. That configuration allows the boot to give the wearer a boost at their ankle during walking, and the idea is that it could potentially be helpful to soldiers who carry heavy loads near the frontlines. The device is equipped with an onboard computer that uses mechanical sensor input and machine learning to constantly try to estimate precisely what the human it’s on is doing and where they are in their walking pattern to give appropriately timed assistance.
So it’s not controlled by its wearer but moves to match their step with a mechanical boost.
“[It] learns each new user’s unique walking pattern so that assistance can be tailored for that user,” Bradford confirmed, adding that the ExoBoot “is then able to adapt to things such as changes in walking speed.”
Researchers recruited 20 test subjects—including multiple soldiers—to participate in the study.
Those involved were at the laboratory for several hours, ARL noted in its post. Sensors were placed on participants from head-to-toe. More than 40 reflective dots were assembled on different parts of their bodies to measure how they move. And 128 dual-layer electroencephalography (or EEG) electrodes developed within the Cognition and Neuroergonomics Collaborative Technology Alliance were put on participants’ scalps via a tool “that looks like a swim cap to measure brain signals.” Those sensors can enable high-fidelity neuroimaging during captured walks.
Study participants stepped on a treadmill inside the Soldier Performance and Equipment Advanced Research Facility, for about an hour with and without the smart device.
“The boot felt natural to the person wearing it. There was no noticeable difference in walking when the ExoBoot was not activated,” Sgt. Maj. Blum noted. “When the ExoBoot was active, it felt like there was a spring in your step.”
As the wearers’ foot transitioned from a forward position to the rear, the ExoBoot would activate, propelling it forward.
“While it felt a little strange at first, within just a few minutes, I was accustomed to the device,” Blum said, drawing on his own experience in the study. “There was a noticeable change to the level of effort required while the device was on and off.”
Pushing Human-Exo Performance
In the coming months, the Army research team is set to analyze the data they collected—and present it at national conferences for feedback. Before that can happen, they’ll need time to understand the neural signals they recorded because that ability to document brain activity with a smart boot remains relatively new. A longer-term aim, according to Bradford, involves pinpointing biometrics that can be integrated into exoskeleton controls.
“While the current version of the ExoBoot is somewhat able to adapt to each new user, not all users experience benefit from the device,” she explained. “Best case scenario is we are able to identify robust signals from the human that can predict the performance of the human-exo team.”
Those signals could in turn act as the basis for novel objective functions that could be fused in the boot’s controls.
“The goal here would be improving human-exo team performance for all users,” Bradford said.
The effort is just one portion of a broader ARL research focus area uniting people and artificial intelligence, to ultimately train teams that can operate together in complex environments. Bradford and Blum additionally both view this work as a prime opportunity for scientists and soldiers to connect.
“It is imperative that we involve soldiers early on in the research process. We need the operational expertise that our soldiers can provide to the development of new technologies,” Blum said. “This is referred to as soldier-centered design.”
He noted that the approach allows the military to garner immediate feedback regarding the applicability of the tools and gauge what capability improvements it might provide.
“The feedback is provided to scientists across the Army research ecosystem, and to academia, industry, and acquisition partners to speed the development of cost-effective, relevant technologies for the warfighter,” Blum said.