摘 要
随着智能交通系统的快速发展，自动驾驶技术已成为研究热点，其中基于视觉感知的路径规划技术在复杂动态环境中具有重要应用价值。本研究旨在通过融合多源视觉信息与先进算法，提升自动驾驶车辆在非结构化场景中的路径规划能力。研究采用深度学习与传统计算机视觉相结合的方法，构建了一种高效的视觉特征提取模型，并结合强化学习优化路径决策过程。实验结果表明，所提出的方法能够在多种复杂路况下实现高精度、低延迟的路径规划，显著提高了车辆行驶的安全性和效率。此外，本研究创新性地引入了场景语义理解模块，增强了系统对动态障碍物的识别能力和实时响应速度。这一成果为自动驾驶技术的实际应用提供了新的思路和解决方案，同时为未来智能交通系统的进一步发展奠定了理论和技术基础。
关键词：自动驾驶路径规划；视觉感知；深度学习
Abstract
With the rapid development of intelligent transportation systems, autonomous driving technology has become a research hotspot, and vision-based path planning techniques play a crucial role in complex dynamic environments. This study aims to enhance the path planning capabilities of autonomous vehicles in unstructured scenarios by integrating multisource visual information with advanced algorithms. A hybrid approach combining deep learning and traditional computer vision was employed to develop an efficient visual feature extraction model, which is further coupled with reinforcement learning to optimize the path decision-making process. Experimental results demonstrate that the proposed method achieves high-precision and low-latency path planning under various complex road conditions, significantly improving the safety and efficiency of vehicle operations. Additionally, this study innovatively incorporates a scene semantic understanding module, thereby enhancing the system's ability to identify dynamic obstacles and its real-time response speed. These achievements provide new insights and solutions for the practical application of autonomous driving technology while laying a theoretical and technical foundation for the further development of future intelligent transportation systems.
Keywords: Autonomous Driving Path Planning；Visual Perception；Deep Learning
目 录
引言 1
一、视觉感知技术基础 1
（一）视觉传感器原理分析 1
（二）图像处理关键技术 2
（三）数据融合方法研究 2
二、路径规划理论框架 2
（一）路径规划基本概念 2
（二）视觉信息的提取与建模 3
（三）动态环境下的路径优化 3
三、视觉感知与路径规划融合 4
（一）视觉数据驱动的决策机制 4
（二）实时路径规划算法设计 4
（三）融合系统的性能评估 5
四、实验验证与结果分析 5
（一）实验平台搭建与测试 5
（二）不同场景下的表现分析 5
（三）系统鲁棒性与局限性讨论 6
结 论 6
致 谢 8
参考文献 9
