Researchers at MIT have developed a new technology, which they have dubbed “RF-Pose.” With RF-Pose, you can, in a sense, “see through” walls — the technology can identify the person, and their actions. This includes whether they are walking, how fast they are moving, waving, sitting, or standing. It integrates artificial intelligence, and offers somewhere around an 83 percent accuracy in identifying people from “a known group.”
While this sounds like an Orwellian nightmare come true, MIT researchers say that the practical applications for the technology could do a lot of good. They say it could be helpful in many sectors, such as law enforcement and healthcare. The leader of the research group, Dina Katabi, elaborated in a statement as reported by NBC News.
“We’ve seen that monitoring patients’ walking speed and ability to do basic activities on their own gives health care providers a window into their lives that they didn’t have before, which could be meaningful for a whole range of diseases.”
Some examples include allowing relatives to monitor their older relatives to find out immediately if they fall, or monitoring someone for signs of Parkinson’s by checking on their gait. Others point out that the technology could be used to develop new types of video games, or used in search and rescue missions, described MIT News.
The technology harnesses radio transmitters, which are reflected by human bodies and later translated by artificial intelligence into “moving stick figures on a screen.” The technology can be trained using photographs of people.
MIT researchers reportedly gathered thousands of images of people doing all sorts of physical activities using a wireless device and a camera to also train the technology. The A.I. eventually was able to “estimate” someone’s movements with no cameras, only using the radio transmissions that were reflected off the person’s body.
Right now, the biggest barrier is that the technology does not work on thick walls. Additionally, there are some huge privacy concerns.
Ginés Hidalgo, a researcher that’s not working on the project, pointed out that “If this camera can be hidden behind or even inside any object, I would never be able to know when I am being monitored.”
On the flip side, researchers say that whenever the technology is installed, that a person would need to undergo a “consent mechanism,” a series of movements to essentially activate or train the technology to recognize them.
#AI at its finest: @MIT #Scientists have developed an #ArtificialIntelligence system, RF Pose, that can detect & correct #postures of #people, even if a solid #wall were to come in between. Know more at #Metamorphoses18 by @CPR_India @NITIAayog @iicdelhi https://t.co/G3KbJIPqOx pic.twitter.com/UG4j6fVBke
— Metamorphoses (@Metamorphoses18) June 25, 2018
Future iterations of this technology could potentially pick up on “micromovements.” An example that is given includes detecting an old person’s hands shaking, which could then alert someone that the person needs to go to the doctor.