Scholarlink

<p>This research develops autonomous electric vehicles for environmentally friendly transportation in urban areas. Utilizing technologies like LiDAR, omnidirectional, and thermal cameras, the vehicles accurately map environments and detect obstacles, enhancing safety. Deep learning methods optimize control systems for efficient autonomous maneuvers. DASH7, Wi-SUN, or Weightless communication systems transmit high-quality images for navigation and obstacle detection. Super-Resolution Convolutional Neural Networks (SRCNN) enhance image resolution, addressing bandwidth limitations and supporting efficient data transmission. Modifications to the SRCNN architecture, including larger kernel sizes and extended neural network layers, improve image quality and processing speed. Efficient image transmission via DASH7, Wi-SUN, or Weightless standards is achieved through advanced image super-resolution techniques. This approach depends on refining the super-resolution model, with four enhanced models developed from three foundational ones. Models 955 and 915 show a better balance between training and validation, indicating superior generalization. Overfitting is addressed through regularization, hyperparameter tuning, early stopping, and dataset augmentation. While all models initially exhibit high PSNR values, some experience a decline in image quality over time. Models 18210, 18610, 181010, and 1895210 better retain image quality compared to models 915, 935, and 955. Learning curves show decreasing loss and increasing PSNR, with models 18210 and 181010 standing out for strong performance.</p>

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

Enhanced Image Transmission in Autonomous Electric Vehicles Using Advanced Image Super-resolution Techniques

  • 期刊论文

  • 发表日期：

    2026年1月28日

查看全文
  • Anindya Afina Carmelya,
  • Arief Suryadi Satyawan,
  • Galura Muhammad Suranegara,
  • Mokhamamad Mirza Etnisa Haqiqi,
  • Salita Ulitia Prini,
  • Pamungkas Daud

摘要

This research develops autonomous electric vehicles for environmentally friendly transportation in urban areas. Utilizing technologies like LiDAR, omnidirectional, and thermal cameras, the vehicles accurately map environments and detect obstacles, enhancing safety. Deep learning methods optimize control systems for efficient autonomous maneuvers. DASH7, Wi-SUN, or Weightless communication systems transmit high-quality images for navigation and obstacle detection. Super-Resolution Convolutional Neural Networks (SRCNN) enhance image resolution, addressing bandwidth limitations and supporting efficient data transmission. Modifications to the SRCNN architecture, including larger kernel sizes and extended neural network layers, improve image quality and processing speed. Efficient image transmission via DASH7, Wi-SUN, or Weightless standards is achieved through advanced image super-resolution techniques. This approach depends on refining the super-resolution model, with four enhanced models developed from three foundational ones. Models 955 and 915 show a better balance between training and validation, indicating superior generalization. Overfitting is addressed through regularization, hyperparameter tuning, early stopping, and dataset augmentation. While all models initially exhibit high PSNR values, some experience a decline in image quality over time. Models 18210, 18610, 181010, and 1895210 better retain image quality compared to models 915, 935, and 955. Learning curves show decreasing loss and increasing PSNR, with models 18210 and 181010 standing out for strong performance.