MAGNETAR: Multipath-Guided Spatial Posteriors for Transmitter Pose Inference in the Upper Mid-Band
Teaching robots to pinpoint hidden radio transmitters in cluttered rooms
A robot trying to locate a radio transmitter in a typical room faces a difficult problem: the radio signal bounces off walls, making it impossible to know from a single measurement whether the transmitter is in front, behind, or to the side. Researchers developed MAGNETAR, a neural network that solves this by computing all plausible combinations of transmitter location and antenna direction rather than guessing a single answer. The system was trained on simulated 10 GHz radio signals and validated on real robot measurements, outperforming simpler approaches that only estimate position.
Robots searching for transmitters—whether for equipment localization in factories, rescue operations in disaster zones, or detecting interference sources in communication networks—currently struggle in realistic indoor settings where signals bounce unpredictably. By computing the full range of possibilities instead of one best guess, MAGNETAR gives robots a more honest picture of their uncertainty, allowing them to make smarter search decisions and requiring fewer measurements to pin down a target's actual location and orientation.