MIT’s RF-Capture Software May Change Sexy Times Forever
- It can know who the person behind a wall is.
- It can trace a person’s handwriting in air from behind a wall.
- It can determine how a person behind a wall is moving .
When it comes to Wi-Fi networks, “I know who you are, and I know what you did” may soon be not so much an ominous accusation as a simple statement of fact.
The MIT Computer Science and Artificial Intelligence Lab’s experimental RF-Capture technology has allowed the researchers to silhouette individuals standing solid walls using wireless signals, but that’s really only the warm-up for this application. The same technique can also track a human hand’s motion and repeated measurements can even train its system to differentiate and identitfy concealed individuals based on differences in their silhouettes. Since first showing off their software’s ability to detect human motion by way of Wi-Fi signal disturbances in 2013, the team has developed the technique over the intervening two years to render basic body shapes from bounced signal reflections.
The concept is so relatively simple, it’s actually somewhat of a wonder it hasn’t been perfected sooner. The research team trims the random noise interference by capturing a series of data frames. Wireless signals transmitted on one side of a wall pass through it simply enough. A body on the other side reflects them for a device to capture and clean up through its software and some upscale processing power. The eventual higher-level data models combine information from numerous captures to construct the most accurate possible silhouettes.
Complex algorithms then parse the collected data to detect shapes and features they’ve been trained to recognize as those of a human body. As it processes scan after scan, the system catalogues what it can and cannot definitely define as a human body part and where each shape fits. As the algorithms recognize distinctive heights, shoulder widths and other distinguishing characteristics, they can then discern one figure behind the wall from another.
So far, the software can accurately tell the differences between 15 different people with about 90-percent accuracy and trace a moving hand through the air accurate to within an inch.
Here are the two edges of the sword. The software could suit positive applications uses from the obvious security and surveilance uses to a currently in-development device that will automatically call 911 if a regular scan of a residence reveals that, say, an elderly grandparent has fallen and can’t get up. The obvious downside being, as the software is tuned to finer and finer degrees of resolution and accuracy, privacy becomes an awkward dimension even with the encryption of any RF-Capture device’s router. That’s why the MIT team is also working toward preventing tracking from unwanted, unauthorized devices and regulations dictating the software’s approved uses.