Australian experts have claimed they’ve identified the answer to preventing smart home devices from spying on their owners.
Australian Centre for Robotics researchers from the University of Sydney, as well as Queensland University of Technology Centre for Robotics (QCR) researchers have created a new privacy-preserving camera to use in devices like autonomous vacuum cleaners, delivery drones and smart fridges.
A creation like this was required after a camera inside an iRobot Roomba J7 robot vacuum took pictures of a woman while she was on the toilet in late 2022. These pictures found their way onto social media.
The images of the woman on the toilet are thought to have been leaked to social media by workers at an AI company, whose job was to label items in footage to train AI systems.
These types of vacuums contain cameras in order to navigate the environment around them and recognize items in a home, including vases, shoes, and chairs, that must be avoided.
There have also been cases of cameras inside baby monitors and smart fridges being hacked or misused by abusers who want to watch their victims.
This new camera works on the premise that most devices don’t need to see as much detail as regular cameras.
Most only use a narrow visual scope, including colour or pattern recognition, which allows them to navigate, assess environmental changes, and recognise objects.
This means visual information can be significantly reduced or distorted that provides everything the robot requires while preserving anonymity for its users.
Global researchers have been attempting to tackle the privacy problem and have focused on scrambling and encoding the images digitally within the device’s in-built computer. This, however, still leaves the original images vulnerable to hackers.
This new creation redesigns optics and analogue electronics of the camera, making sure it only collects specific information that is required to perform the task at hand.
The circuits will collect statistics on the lightness and darkness of various points and convert it to a unique “fingerprint.” This gets sent to the computer of the device to be read. Full images will never be received by the camera, preventing leakage, misuse, and hacking.
Leader of the project at the Australian Centre for Robotics, Dr Don Dansereau said, “So much of the approach to the technology today is about upgrading digital capacity so we can collect more and more data. But when it comes to privacy, there is a strong argument for returning to fixed analogue components that put hard limits on the amount of data that can be captured.”
If this new camera pans out, it could play a large role in the growth of the so-called internet of things, especially when it comes to the growing global focus on consumers’ right to privacy.
Currently, these researchers are exploring industry partners to look into manufacturing at scale. They hope these cameras will make it to market within the next two-to-three years. They also believe autonomous flight applications, such as delivery drones, could be a potential market for these cameras.
