According to a team of researchers led by a University of Maryland computer scientist, those cute little autonomous floor cleaners could be sweeping up more than just dust in your home. The researchers demonstrated how popular robotic vacuums can be remotely hacked to collect audio, despite the fact that the devices were never designed to do so.
The researchers, led by Assistant Professor Nirupam Roy, acquired data from a popular vacuum robot’s laser-based navigation system and used signal processing and deep learning methods to recover voice and recognize television shows playing in the same room as the device.
The research shows how any device that employs light detection and ranging (Lidar) technology may be used to gather sound. This work, which Roy co-led with Assistant Professor Jun Han at the National University of Singapore was presented today at the Association for Computing Machinery’s Conference on Embedded Networked Sensor Systems.
“We welcome these things into our homes and don’t think twice about it,” Roy, who is also affiliated with the University of Maryland Center for Advanced Computing Studies, said (UMIACS). “But we have shown that even though these devices don’t have microphones, we can repurpose the systems they use for navigation to spy on conversations and potentially reveal private information.”
According to privacy experts, the home maps created by the vacuum bots utilizing Lidar, enabling them to avoid collisions and possible privacy violations. The maps, which are often saved in the cloud, might provide marketers with information such as house size, which signals income level, and other lifestyle-related information. Roy and his colleagues asked whether the Lidar in these robots may offer security problems as sound recording devices in users’ homes or companies.
Sound waves cause objects to vibrate, and these vibrations cause slight variations in the light bouncing off an object. Laser microphones, used in espionage since the 1940s, are capable of converting those variations back into sound waves. Laser microphones, on the other hand, depend on a focused laser beam bouncing off exceptionally smooth surfaces, such as glass windows.
A vacuum Lidar, on the other hand, uses a laser to scan the surroundings and detect light reflected back by irregularly shaped and dense objects. The dispersed signal acquired by the vacuum sensor contains just a small portion of the information required to recover sound waves. The researchers were initially unsure if a vacuum bot’s Lidar system could be manipulated to function as a microphone, and if the resulting signal could be meaningfully interpreted..
Initially, they hacked a robot vacuum to demonstrate that they could alter the laser beam’s location and communicate the data to their computers through Wi-Fi without interfering with the device’s navigation.
They then experimented with two different sound sources: a human person repeating numbers via computer speakers and sounds from various television episodes played through a TV sound bar. Roy and his colleagues then captured the laser signal sensed by the vacuum’s navigation system as it bounced off a variety of objects—including a trash can, a cardboard box, a takeout container and a plastic bag—placed near the sound source.
The researchers ran the generated signals through deep learning algorithms that had been trained to recognize musical sequences from television episodes or to match human voices. Its computer technology, LidarPhone, recognized and matched spoken numbers with 90% accuracy. It also recognized the programs with more than 90% accuracy from a minute’s worth of recording.
“When you think that we are all purchasing meals over the phone and having meetings over the web, and we are often speaking our credit card or bank information,” Roy added.
“But what is even more concerning for me is that it can reveal much more personal information” including lifestyle, work hours and political affiliation, he said. “This is critical for someone who may wish to affect political elections or send me extremely targeted messages.”
One example of a possible vulnerability to Lidar-based eavesdropping is robot vacuum cleaners. Many other devices could be open to similar attacks such as smartphone infrared sensors used for face recognition or passive infrared sensors used for motion detection, the researchers said.
“I think this is crucial work that will raise manufacturers’ awareness of these threats and prompt the security and privacy communities to develop ways to avoid these kind of assaults,” Roy added.
Related Questions
-
Can robot vacuums listen to you?
You should be concerned about more than simply speakers. Any connected device in your house can be used to monitor you. This includes your robot vacuum cleaner.
-
Does Roomba Track your house?
Roomba can also learn about your house while cleaning it. It typically takes three (3) to five (5) cleaning missions or Mapping Runs to generate a fully developed Imprint™ Smart Map that you can then customize and use.
-
Does my shark robot vacuum have a camera?
It has simple controls, an on-board camera, and integrated technology to essentially provide a “hands-off” cleaning experience. This Shark robot vacuum cleaner is intended to: Travel in straighter lines. Clean many kinds of floors.
-
Can my downstairs neighbors hear my Roomba?
A: I can tell you from experience the Roomba can be heard if your walls are thin and you share floorboards with your neighbor. It bangs against baseboards and furniture legs, and if you just have hardwood floors, your neighbors will hear it.