The Machines now have X-ray vision. A new piece of software has been trained to use wifi signals — which pass through walls, but bounce off living tissue — to monitor the movements, breathing, and heartbeats of humans on the other side of those walls. The researchers say this new tech’s promise lies in areas like remote healthcare, particularly elder care, but it’s hard to ignore slightly more dystopian applications.

While it’s easy to think of this new technology as a futuristic Life Alert® monitor, it’s worth noting that at least one member of the research team at the Massachusetts Institute of Technology behind the innovation has previously received funding from the Pentagon’s Defense Advanced Research Projects Agency (DARPA). Another also presented work at a security research symposium curated by a c-suite member of In-Q-Tel, the CIA’s high-tech venture capital firm.

Inverse recently caught up with project’s leader Dina Katabi, a 2013 MacArthur “Genius Grant” Fellow who teaches electrical engineering and computer science at MIT, to talk about how the new tech may be used.

“We actually are tracking 14 different joints on the body … the head, the neck, the shoulders, the elbows, the wrists, the hips, the knees, and the feet,” Katabi said. “So you can get the full stick-figure that is dynamically moving with the individuals that are obstructed from you — and that’s something new that was not possible before.”

RF-Pose A.I. using turning machine learning and a wifi signal into X-ray vision
An animation created by the RF-Pose software as it translates a wifi signal into a visual of human motion behind a wall.

The technology works a little bit like radar, but to teach their neural network how to interpret these granular bits of human activity, the team at MIT’s Computer Science and Artificial Intelligence Laboratory (CSAIL) had to create two separate A.I.s: a student and a teacher.

Many of the A.I.s designed to identify data, according to Katabi, are first trained by hand by humans. Developers might feed a machine-learning algorithm hundreds of labelled photos of a jelly doughnut, for example, until the A.I. becomes perfectly adept at picking out jelly doughnut photos all by itself.

three doughnuts

The problem is, identifying human activity from wifi signals isn’t really something that even humans know how to do themselves. So the team developed one A.I. program that monitored human movements with a camera, on one side of a wall, and fed that information to their wifi X-ray A.I., called RF-Pose, as it struggled to make sense of the radio waves passing through that wall on the other side.

The research builds off of a longstanding project at CSAIL lead by Katabi, which hopes to use this wifi tracking to help passively monitor the elderly and automate any emergency alerts to EMTs and medical professionals if they were to fall or suffer some other injury. (The state of the art in this field right now remains the cheesey Life Alert® dongle of internet fame, which the world’s grandmas and grandpas have to press themselves to notify first responders of their situation.)

Katabi and two of her students presented a nascent version of their elder care technology built around this research, Emerald, during President Obama’s first-ever White House Demo Day in August of 2015.

Then-President Barack Obama meets with MIT professor Dina Katabi and two members of her Emerald project in 2015

“So, if you look at our work that we presented to Obama,” Katabi said in a phone interview, “we could track the person using wireless technology — we could even track them through walls and occlusions — but you would just see a blob that moves with that person.”

Katabi says she hopes to get the RF-Pose A.I. sophisticated enough that it can help monitor a variety of human health data tied to movement, identifying the early manifestations and progression of diseases like Parkinson’s or multiple sclerosis (MS). (Prior versions of this research could already track physiological data like breathing patterns and heart rate.) She also said RF-Pose’s underlying tech could easily apply to a number of other potential uses: from search-and-rescue missions retrieving avalanche victims, to wild futuristic revivals of Xbox Kinect, to intervening in dicey hostage situations between terrorists and law enforcement.

Like many teams that are capable of such advanced computer science research, plenty of members of the RF-Pose team are also no strangers to funding from the U.S. government’s sprawling national security sector.

One of the more senior co-authors of this research, Antonio Torralba of MIT’s Computer Science and Artificial Intelligence Laboratory, is slated for presentation next week at the Conference on Computer Vision and Pattern Recognition, and has a history of DARPA funding. His work on a “context-based vision system for place and object recognition” was sponsored by the Air Force and the U.S. Department of Interior-Fort Huachuca under DARPA/MARS in 2003. He also spoke on object recognition for DARPA’s GAME Workshop in 2006. And, in 2011, he contributed to a DARPA-funded effort to develop a massive benchmark dataset for automated “event recognition” in surveillance video.

As for Katabi, her computer science research has also occasionally brought her into close proximity with the American government’s covert project research. She presented research on delivering “tamper-evident” digital packets of information designed to foil “man in the middle” hacking attacks to the 20th USENIX Security Symposiumin 2011. The week-long conference’s four-person “invited talks committee” included Dan Geer, the chief information security officer of the CIA’s financial leash on Silicon Valley innovation, In-Q-Tel.

“We very much realize the technology can be misused to invade privacy and all these other issues,” Katabi said when asked about the possibility of NSA-style surveillance applications. “On our side, we always try to develop counter-technologies to develop mechanisms that would make sure — if someone was trying to use the technology against you — you have other technologies that you can use to counter it.”

She also said that there needs to be more consistent solutions at the policy level to deal with advances in technology that protects citizens and their privacy without stifling innovation.

“We do have IP [rights] and we try to protect the technology from people who want to use it in something that we don’t agree with,” Katabi said, adding, “Let me tell you: When somebody wants to abuse the technology, they’re not going to worry about IP rights and intellectual property. They’re gonna abuse everything.”