HoloRadar: Robot "looks" around corners using radio waves and AI
When robots can "see" areas not optically visible, they can detect dangers early. HoloRadar, based on radio waves, aims to make this possible.
Researchers from the University of Pennsylvania with a robot that can "look" around corners using HoloRadar.
(Image: Sylvia Zhang/Penn Engineering)
An engineering team from the University of Pennsylvania has developed a system that allows robots to “look” around corners using radio waves and artificial intelligence (AI). The system, called HoloRadar, is intended to enable robots to orient themselves better in cluttered indoor environments and in poor lighting conditions. In industrial settings, for example, this is intended to increase safety.
The HoloRadar system uses long-wavelength radio as its basis. Walls and other flat surfaces can reflect these long-wavelength radio waves better, effectively turning them into mirrors. Walls, floors, and ceilings reflect information about hidden areas through the radio waves, which is then analyzed by an AI and reconstructed as a 3D view. In a yet-to-be-published study, which the researchers presented at the 39th annual Conference on Neural Information Processing Systems (NeurIPS), the results are summarized.
Empfohlener redaktioneller Inhalt
Mit Ihrer Zustimmung wird hier ein externes YouTube-Video (Google Ireland Limited) geladen.
Ich bin damit einverstanden, dass mir externe Inhalte angezeigt werden. Damit können personenbezogene Daten an Drittplattformen (Google Ireland Limited) übermittelt werden. Mehr dazu in unserer Datenschutzerklärung.
The AI works in two stages: in the first stage, it improves the resolution of the raw radio signals and detects backscatter generated by multiple reflections along different reflection paths. In a second stage, the reflections are traced back using a physical model. This allows the environmental reflection effects to be canceled out and the real scene to be reconstructed.
“In a way, the challenge is similar to entering a room full of mirrors,” says Zitong Lan, a doctoral candidate in electrical and systems engineering and co-author of the study. “You see countless copies of the same object reflected in different places, and the difficulty is figuring out where things actually are. Our system learns to reverse this process in a physically grounded way.”
“Seeing” Around Corners
The technology can be used by robots to “see” areas that are not optically visible to them. This allows robots to detect people in warehouses, for example, even if they are moving outside the visible range, such as around a corner.
Similar systems have existed before, the scientists explain. However, they mostly used visible light. Shadows and indirect reflections are used to reconstruct objects. Such systems, however, are highly dependent on existing lighting conditions.
Videos by heise
HoloRadar, on the other hand, works independently of lighting conditions. “This system is mobile, operates in real-time, and requires no controlled lighting,” says Mingmin Zhao, a professor of computer and information science at the University of Pennsylvania.
However, the technology cannot completely replace existing sensors for environmental perception in robots and autonomous vehicles. HoloRadar is intended as a supplement to Lidar, Radar, and camera systems, as it makes objects and people outside the visible range visible.
(olb)
