Scholarlink

<p>Recent advancements in Deep Learning (DL), particularly convolutional neural networks and transformer-based architectures, have substantially improved perception and object detection capabilities in autonomous vehicles (AVs). Yet, deploying these methods in real-world driving remains challenging due to computational limitations, sensor heterogeneity, and highly dynamic environments. This review provides an integrative examination of state-of-the-art perception systems, emphasizing sensor fusion strategies and DL-based object detection frameworks. Unlike previous surveys, this work offers a unified perspective that bridges multi-sensor data fusion, deep learning perception, and the emerging role of Large Language Models (LLMs). By critically analyzing key methodologies, we highlight their practical strengths, limitations, and interdependencies. Furthermore, we discuss how LLMs can introduce contextual reasoning and adaptive understanding into AV perception pipelines, paving the way for more robust, intelligent, and interpretable autonomous driving systems.</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

Perception in autonomous vehicles: catalyzing evolution in the automotive landscape

  • 期刊论文

  • 发表日期：

    2026年7月9日

查看全文
  • Ihssane Bouasria,
  • Walid Jebrane,
  • Nabil El Akchioui

摘要

Recent advancements in Deep Learning (DL), particularly convolutional neural networks and transformer-based architectures, have substantially improved perception and object detection capabilities in autonomous vehicles (AVs). Yet, deploying these methods in real-world driving remains challenging due to computational limitations, sensor heterogeneity, and highly dynamic environments. This review provides an integrative examination of state-of-the-art perception systems, emphasizing sensor fusion strategies and DL-based object detection frameworks. Unlike previous surveys, this work offers a unified perspective that bridges multi-sensor data fusion, deep learning perception, and the emerging role of Large Language Models (LLMs). By critically analyzing key methodologies, we highlight their practical strengths, limitations, and interdependencies. Furthermore, we discuss how LLMs can introduce contextual reasoning and adaptive understanding into AV perception pipelines, paving the way for more robust, intelligent, and interpretable autonomous driving systems.