The Massachusetts Institute of Technology has created technology that allows artificial intelligence to identify objects inside closed boxes, drawers, or behind walls, using the same waves as Wi-Fi.
The Massachusetts Institute of Technology has created technology that allows artificial intelligence to identify objects inside closed boxes, drawers, or behind walls, using the same waves as Wi-Fi.
MIT researchers have presented the mmNorm method. This is a new technique that uses millimeter waves (in the Wi-Fi range) to 3D reconstruct objects invisible to the human eye. According to TechXplore, the system achieves 96% accuracy in recognizing more than 60 objects, including cups, spoons, or tools hidden in boxes.
Unlike previous methods that used low-resolution image projection, mmNorm works with the phenomenon of specular reflection of millimeter waves. Instead of just recording where the signal is coming from, the system analyzes the direction of the surface from which it is reflected. This allows it to accurately determine the shape and curvature of hidden objects.
Signals are received from multiple antennas. Each of them «votes» for a certain orientation of the surface, and the system combines this data into a single model, the Signal Kinetics laboratory explains.
The technology has shown high efficiency on objects made of wood, glass, plastic, and rubber, although it still has difficulties with metal or very thick materials.
The mmNorm method could be used in warehouse logistics, robotic search and rescue operations, and security screening, including non-contact analysis of the contents of bags or boxes. This is MIT’s second attempt in the past year to radically rethink how robots can «see» the world without the use of optics.
We previously wrote about how Australian researchers trained real beetles to carry out rescue missions. Thanks to a microchip on their backs, the insects can be controlled remotely, even made to climb walls in disaster zones.
