Embodied Flight Perception via Forewing‐Hindwing Integration in Lepidopteran‐Inspired Aerial Robotics

Abstract Flapping‐wing robots (FWRs) struggle to achieve the agility of insects, largely because they lack the sophisticated sensing capabilities of their biological counterparts. Butterflies and moths, for instance, use their wings as sensory organs, with distributed sensors on both forewings and hindwings working together to provide a robust perception of flight states. Here, this study presents a bioinspired aerial robot that replicates this integrated forewing‐hindwing sensing principle. A flapping‐wing platform is developed with artificially veined, flexible wings and a distributed network of piezoelectric sensors. A core finding of this work is that fusing sensor data from both forewings and hindwings enables a neural network to accurately estimate flight parameters (pitch, stroke angle, and frequency) with significantly higher accuracy than using either wing alone. This multi‐pathway fusion also allows the system to perfectly classify complex flight maneuvers (takeoff, hovering, ascent, and descent). Finally, the real‐world utility of this embodied perception system is developed by successfully deploying it on a free‐flying robot, enabling accurate flight parameter estimation. This work establishes a practical framework for building agile, autonomous FWRs that perceive the world through their wings, much like insects do.