SNOCHIP INC — Department of Defense STTR Phase I: N23A-T023

Description

This proposal aims to develop a low-power multifunctional imager that can simultaneously capture both near-field and far-field images. The proposed system consists of three essential elements: a multi-functional metalens, a vision algorithm for 3D estimation, object counting and tracking, and a low-power camera sensor, enabling rapid analysis, interpretation, and transmission of the collected data via satellite communications (e.g., iridium, Starlink, etc.).   The multifunctional metalens will be designed to image objects in both the near-field (order 1 cm) and mid-field (order 1 m). Furthermore, the metalens will capture one sharply focused and one slightly blurred image of near-field targets on the low-power camera sensor, which will allow for extracting depth information for three-dimensional tracking and imaging. An efficient and fast computer vision algorithm will be developed to process the images captured by the metalens. Additionally, the algorithm will include a computationally efficient neural network architecture to detect and count marine creatures in the mid-field image. To achieve this, we will utilize an ultra-low power 60 FPS, VGA resolution (640×480) camera sensor equipped with an AI accelerator-embedded ASIC processor. The sensor can be powered with only 1-2 to 2.2V power sources, and its processor supports programs written using TensorFlow Lite, including a wide range of data processing operations required by the proposed algorithm. The sensor board will be connected to the ultra-low power Swarm M138 Satellite Transceiver. Various image compression algorithms will be studied to efficiently transmit data.   In Phase I, we will develop the required algorithms for 3D imaging and communication between the sensor and the satellite transceiver while the metalens' design and fabrication are undergoing. Cleanroom tools at Princeton University will be used for metalens fabrication. In Phase I Option, we will assemble the sensor and the metalens as an imager and test its performance in a lab environment.   Our unique low-power multifunctional imager has significant potential to improve our understanding of the physical, chemical, and biological properties of the underwater marine environment. This technology could be beneficial for a range of applications, including environmental monitoring, oceanography, aquaculture, and marine biology. The imager’s ability to operate both below and above the water surface and to provide rapid analysis, interpretation, and transmission of collected data via satellite communications makes it a valuable tool for underwater research and environmental monitoring.