Unmanned driving road edge detection method and device in unstructured scene

By constructing an unstructured road edge detection dataset and a Transformer-based deep network model, the accuracy and stability issues of road edge detection in unstructured roads were solved, achieving centimeter-level detection accuracy and supporting safe driving and path planning for autonomous vehicles.

CN122415662APending Publication Date: 2026-07-17
View PDF 0 Cites 0 Cited by

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Filing Date
2026-04-22
Publication Date
2026-07-17

Smart Images

  • Figure CN122415662A_ABST
    Figure CN122415662A_ABST
Patent Text Reader

Abstract

本发明提供一种非结构化场景下的无人驾驶道路边沿检测方法及设备，属于无人车技术领域，该方法包括对激光雷达点云序列进行鸟瞰图（Bird Eye’s View，BEV）特征提取，并利用可变形注意力机制增强BEV特征；设计一种结合不同角度与曲率的锚框先验，迭代回归预测路边沿点在三维空间中的位置坐标；通过帧间时序增强与双向注意力融合，有效解决因车辆、行人等动态障碍物遮挡导致的路边沿信息缺失与不连续问题。本发明解决了非结构化道路边沿因缺乏规则几何特征、边界模糊及环境动态变化所导致的检测难题，能够在少量标注数据的训练下实现厘米级的检测精度，为无人驾驶车辆的贴边行驶、安全避障与路径规划提供有效支撑。
Need to check novelty before this filing date? Find Prior Art