Autonomous Intelligent Metasurface for Wireless Communications and Contactless Human Sensing

ABSTRACT Metamaterials and metasurfaces have revolutionized the electromagnetic (EM) wave control, and programmable metasurfaces enabled dynamic manipulation of wavefronts. However, the current programmable metasurfaces lack autonomous intelligence and rely heavily on manual control or pre‐specified rule sets, limiting the self‐adaptability in dynamic scenarios. Here, we propose an autonomous intelligent metasurface (AIM) that seamlessly integrates the programmable metasurface with the large language model (LLM)‐based reasoning, multimodal control, and closed‐loop feedback. Inspired by human cognition, AIM is structured into six functional modules to mimic the ear, brain, eye, hand, neuron, and mouth, which enable natural language interaction, environmental understanding, and autonomous EM manipulation. Notably, AIM supports integrated sensing and communication using widely‐used WiFi signals, realizing simultaneous data transmissions and real‐time human vital‐sign monitoring without modifying the communication protocols. Demonstrated in smart indoor settings, AIM can provide us with a transformative framework for intelligent and user‐driven EM interaction, with broad potential for future applications in smart homes, healthcare, and human‐machine interfaces.