Electromagnetic (EM) sensing is a wide-spread contactless examination technique in science, engineering and military. However, conventional sensing systems are mostly lack of intelligence, which not only require expensive hardware and complicated computational algorithms, but also pose important challenges for advanced in-situ sensing. To address this shortcoming, we propose the concept of intelligent sensing by designing a programmable metasurface for data-driven learnable data acquisition, and integrating it into a data-driven learnable data processing pipeline. This strategy allows to learn an optimal sensing chain in systematic sense of variational autoencoder, i.e., to jointly learn an optimal measurement strategy along with matching data post processing schemes. A three-port deep artificial neural network (ANN) is designed to characterize the measurement process, such that an optimal measurement strategy is adaptive to the subject of interest by controlling the programmable metasurface for manipulating the EM illuminations. We design and fabricate a proof-of-principle sensing system in microwave, and demonstrate experimentally its significance on the high-quality imaging and high-accuracy object recognition from a remarkably reduced number of measurements. We faithfully expect that the presented methodology will provide us with a fundamentally new perspective on the design of intelligent sensing architectures at various frequencies, and beyond.
Download