CMU Researchers Develop Tech to See People Through Walls

X-ray vision is a thing of fantasy and science fiction, which you’ve probably seen in fictional characters like Superman. However, efforts have been made to make seeing through walls, particularly seeing people, a reality. This would open up great possibilities for more accessible human tracking, beneficial, for instance, to military personnel or to individuals who require extra surveillance at home or at the office.

Several years back, MIT researchers tried using radio waves and cell phone signals coupled with AI to make the superhuman ability happen. Despite their great efforts, however, the algorithm they built only made images of low resolution come up, providing the visibility of humans but not their identification, incompatible with most of its potential use cases.

Scientists and researchers from Carnegie Mellon University (CMU) have found a way to create an improved human tracking tech, which tracks and maps key points of the human body using WiFi signals, identifying them with AI and computer vision algorithms.

Human Tracking Methods

Human tracking tech and algorithms, especially for military use, have been under development and use for years. Commonly obtained using RGB cameras or advanced, specialized hardware such as LiDAR and radars, human pose and identification estimation require plenty of power and costs.

CMU’s team of researchers developed their WiFi-based surveillance tech to turn human tracking technology into a more affordable and more accessible craft. In their recently published paper, it is elaborated that the team uses DensePose, a deep neural network developed by researchers from Imperial College London, University College London, and Facebook AI.

The network maps the phase and amplitude of WiFi signals, taking that information as input to be translated by UV coordinates. The signal sends about 24 key points of the human body, such as the head, arms, and legs, to DensePose to be analyzed using computer vision algorithms. This then results in having a 3D model’s surface being projected to a 2D image, done to map computer-generated images.

This way, instead of obtaining maps from RGB cameras or LiDAR, which in the case of the latter would cost $700 each, DensePose allows the research team to accurately map multiple subjects’ locations and poses through walls only using WiFi antennas, with routers that would only cost $30. Aside from the cheap setup, the CMU research team also noted that their WiFi-based tech isn’t at all impacted by poor lighting or obstruction, unlike most existing systems out there.

With how widespread Wi-Fi already is, especially in developed countries, the human tracking technology would then become much easier to scale. The team imagines that individuals can utilize the tech to monitor the well-being of the elderly in their homes, to monitor patients of home healthcare who do not want cameras in the room, or to identify any suspicious and unwanted behaviors.

Privacy Matters

The accessibility of Wi-FI has made the invasion of privacy a cause for concern, raising questions regarding the security implications the human tracking technology will entail. However, the team claims that the tech, especially once available to the public, will be “privacy-friendly,” as it doesn’t provide completely identifiable and clear images of the subjects.

Despite this statement, it is still questionable how safe the human tracking technology can be in the midst of cybersecurity threats, including hacking, still on the rise after the past few years, with several parties still unsure of whether the tech will exclusively be used for good or if it will result in the exploitation of one’s privacy.

If worries regarding safety and privacy issues can be managed, the work done by the CMU research team can lead to human tracking technology being more ubiquitous and accessible in the future without the need for households, institutes, and organizations to spend hundreds and thousands of dollars for required surveillance and monitoring.

Meme & AI-Generated Picture

Job Posting

• Cash App - Staff Data Scientist, Controls/Access - San Francisco, CA+ (Remote/Hybrid)

• Keeper Security, Inc. - QA Automation Specialist - EL Dorado Hills, CA (Remote/Hybrid)

• Motorola Solutions - Sr. Systems Analyst - San Francisco, CA+ (Remote)

• Wowza - Senior Software Development Engineer - Portland, OR+ (Remote)

Promote your product/service to Digger Insights’ Community

Advertise with Digger Insights. Digger Insights’ Miners are professionals and business owners with diverse Industry backgrounds who are looking for interesting and helpful tools, products, services, jobs, events, apps, and books. Email us [email protected]

Your feedback would be greatly appreciated. Send it to [email protected]