Scholarlink

Recent progresses of Si photonic applications in new areas are introduced here. One is integrated solid state Si LiDARLiDAR (light detection and ranging) with optical beam steering arrays for autonomous cars, free-space communication links, etc. After surveying principles of key optical phased array techniques and frequency-modulated continuous wave (FMCW) operation, various FMCW–LiDAR systems developed to date are surveyed together with reviewing several beam steering elements such as high-performance optical phased arrays, focal plane arrays, and slow-light gratings. The other progress is optical neural network (ONN) using an integrated Si photonic chip for deep learning of image recognition and 1-D to 3-D dataset classification. First, basic elements of the ONN with input-/hidden-/output-layers, matrix–vector multiplication, and nonlinear activation function are reviewed. Then, novel various ONN architectures such as real/complex-valued networks, nonlinear mapping-based photonic classifier, and space-efficient diffractive ONN for Fourier transform are reviewed. Latest experiments on inference and training tasksTraining task such as logic gate realization, nonlinear dataset classification, and handwriting recognition are also shown.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Applications of Silicon Photonic Waveguides (II) LiDAR and Optical Neural Network

  • 章节

  • 发表日期：

    2025年

查看全文
  • Kiyoshi Asakawa,
  • Yoshimasa Sugimoto,
  • Shigeru Nakamura

摘要

Recent progresses of Si photonic applications in new areas are introduced here. One is integrated solid state Si LiDARLiDAR (light detection and ranging) with optical beam steering arrays for autonomous cars, free-space communication links, etc. After surveying principles of key optical phased array techniques and frequency-modulated continuous wave (FMCW) operation, various FMCW–LiDAR systems developed to date are surveyed together with reviewing several beam steering elements such as high-performance optical phased arrays, focal plane arrays, and slow-light gratings. The other progress is optical neural network (ONN) using an integrated Si photonic chip for deep learning of image recognition and 1-D to 3-D dataset classification. First, basic elements of the ONN with input-/hidden-/output-layers, matrix–vector multiplication, and nonlinear activation function are reviewed. Then, novel various ONN architectures such as real/complex-valued networks, nonlinear mapping-based photonic classifier, and space-efficient diffractive ONN for Fourier transform are reviewed. Latest experiments on inference and training tasksTraining task such as logic gate realization, nonlinear dataset classification, and handwriting recognition are also shown.