Aberration-corrected three-dimensional positioning with a single-shot metalens array

Three-dimensional (3D) positioning with the correction of imaging aberrations in the photonic platform remains challenging. Here, we combine techniques from nanophotonics and machine vision to significantly improve the imaging and positioning performance. We use a titanium dioxide metalens array operating in the visible region to realize multipole imaging and introduce a cross-correlation-based gradient descent algorithm to analyze the intensity distribution in the image plane. This corrects the monochromatic aberrations to improve the imaging quality. Analysis of the two-dimensional aberration-corrected information in the image plane enables the 3D coordinates of the object to be determined with a measured relative accuracy of 0.60%–1.31%. We also demonstrate the effectiveness of the metalens array for arbitrary incident polarization states. Our approach is single-shot, compact, aberration-corrected, polarization-insensitive, and paves the way for future integrated photonic robotic vision systems and intelligent sensing platforms that are feasible on the submillimeter scale, such as face recognition, autonomous vehicles, microrobots, and wearable intelligent devices.