Bioinspired planar intelligent nanophotonic sensor for wide-angle accurate motion perception and prediction

In nature, certain insects possess specialised compound eye structures that provide an ultra-wide field-of-view (FoV) and rapid response capabilities, enabling them to capture prey and avoid obstacles.Herein, inspired by compound eyes, a planar intelligent nanophotonic sensor (PINS) based on a metalens array, which possesses an ultrawide horizontal FoV exceeding 135°, is demonstrated.By leveraging a deep neural network, meta-motion sense (MMS), accurate optical flow can be extracted from PINS-captured wide-FoV scenes, enabling a comprehensive characterisation of the motion velocities and directions of all dynamic objects.Compared to traditional machine-vision-based object recognition algorithms, the proposed approach exhibits significantly higher accuracy and robustness, particularly in detecting small, slow, or background-blended moving targets, and offers an intelligent predictive capability for forecasting the motion trajectories of objects.The proposed device combines the advantages of high compactness, superior motion-detection performance, and intelligent functionality, offering a promising foundation for next-generation applications in autonomous navigation, situational awareness, and military surveillance.