All-integrated multidimensional optical sensing with a photonic neuromorphic processor

Gu et al., Sci. Adv. 11, eadu7277 (2025) 30 May 2025 S c i e n c e A d va n c e s | R e s e a r c h A r t i c l e 1 of 10 O P T I C S All-­integrated multidimensional optical sensing with a photonic neuromorphic processor Zhijuan Gu1†, Yang Shi1†, Zhangming Zhu1, Zuhang Li1, Mingjie Zou1, Changming Yang1, Yang Liu1,2, Yu Yu1*, Xinliang Zhang1* Multidimensional optical sensing is crucial in information technology and modern intelligent systems. Despite advancement in optical sensing, capturing multidimensional light field information remains challenging, typi- cally implemented using cascaded single-­dimensional sensors and discrete optoelectrical components for infor- mation decoupling. Here, we present an all-­integrated multidimensional sensing chip incorporating a light field sensitizer and a photonic neural network processor. The inverse-­designed sensitizer projects the multidimension- al input into multiple channels; each dimension is then decoupled through the reconfigurable nonlinear neural network. We experimentally achieved 91% high accuracy for single-­shot, concurrent sensing of intensity, polariza- tion, and wavelength using a well-­trained five-­layer neuromorphic system. The fully on-­chip system eliminates optical-­electrical conversion and offline digital processing, enabling low-­latency and high energy efficiency. Moreover, we achieved stabilization and recovery of high-­speed signals at 100 gigabytes per second under ran- domly perturbed polarization and wavelengths. This work shows the potential for low-­latency, energy-­efficient optical sensing and complex information processing using neuromorphic integrated photonics. INTRODUCTION Optical sensing plays an essential role in both physical sciences and our daily lives (1–3), taking the advantages of light fields that can be encoded with rich information with broadband capacity and can be instantaneously transmitted with low loss. The vectorial light fields involve diverse physical properties and dimensions, including inten- sity, polarization, and wavelength (4). The capability of measuring such multidimensional light fields is the key to the detection of abundant encoded information, which can enhance decision analysis and multi- plexed data transmission capacity. Recently, multidimensional sensing attracts increasing attention (5) and exhibits advantages in sensor-­rich domains, such as the satellite remote sensing and biomedical imaging (6, 7). Equally important, the ability to analyze multidimensional light fields can facilitate optical signal processing and enhance optical com- munications (8) where environmental changes induced by polarization disturbance can severely affect the communication quality (9). Multidi- mensional sensing allows for the acquisition of independent dimen- sions of light fields outputs, enabling stable detection of communication signals without interference. Integrat