MEMS Reconfigurable Antenna for Multimodal Perception in Humanoid Robots

Addressing the critical challenge of synergistic integration between multimodal perception (radar imaging + inter-robot communication) and high-rate data transmission for humanoid robots in unstructured environments, this paper proposes a terahertz-band sensing-communication integrated antenna based on MEMS dynamic tuning. Through innovative design of a silicon-based MEMS electrostatic comb-resonator coupled structure, the antenna achieves active reconfigurability within the 290-305 GHz band, demonstrating return loss better than -30 dB and peak gain of 1.68 dBi. To resolve temporal conflicts between sensing and communication, a time-division multiplexing mechanism is developed: During communication cycles (60% duty ratio), it supports 28 Gbps data throughput, while in sensing cycles (40% duty ratio), it enables obstacle detection with ±0.3 mm precision–a resolution sufficient for delicate operations such as minimally invasive surgery where robotic fingers require precise tissue manipulation. This represents the first demonstration of coordinated submillimeter-scale environmental mapping and ultra-low-latency communication within a single antenna. Experimental validation on a bipedal robot integrated with this antenna reveals: Under complex gaits (2 Hz stride frequency, ±5° amplitude), communication interruption rate decreases from 12.7% (conventional antennas) to 0.8%, with obstacle miss-detection rate reduced to 1.2%–meeting dynamic obstacle avoidance specifications in China’s National Standard GB/T 39473-2020 "Safety Requirements for Service Robots". This research establishes hardware foundations for human-robot-environment closed-loop interaction systems.