融合视觉语义信息与激光雷达的四足机器人路径规划方法

By integrating visual semantic information with LiDAR path planning methods, the problem of blind spots in perception for quadruped robots in industrial scenarios has been solved, achieving safer and more efficient navigation and obstacle avoidance capabilities.

CN120628103BActive Publication Date: 2026-07-17SOUTHWEST JIAOTONG UNIV

Patent Information

Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
SOUTHWEST JIAOTONG UNIV
Filing Date
2025-06-06
Publication Date
2026-07-17

AI Technical Summary

Technical Problem

In industrial settings such as railway maintenance, the perception system of quadruped robots is unable to accurately identify the location of ditches, causing them to mistakenly enter the ditches and resulting in accidents such as falls and collisions, which affects navigation safety and work efficiency.

Method used

A path planning method integrating visual semantic information and LiDAR is proposed. By acquiring key ground semantic information and depth images, it is converted into key ground semantic point cloud information of LiDAR, multi-source point cloud fusion is performed, non-ground interference points are filtered out, a semantically enhanced local path planning model is constructed, and a semantic penalty term is introduced to avoid dangerous areas.

Benefits of technology

It significantly improves the navigation safety and task execution efficiency of quadruped robots in high-risk industrial environments, and achieves a more stable dynamic obstacle avoidance strategy.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN120628103B_ABST
    Figure CN120628103B_ABST
Patent Text Reader

Abstract

本申请涉及一种融合视觉语义信息与激光雷达的四足机器人路径规划方法。所述方法包括:获取环境数据,环境数据包括关键地面语义信息、深度图像和原始激光雷达点云信息;基于深度图像将关键地面语义信息转换为激光雷达关键地面语义点云信息,与原始激光雷达点云信息进行空间融合得到统一参考系下的多源点云融合信息;对多源点云融合信息进行预处理,滤除非地面干扰点;基于语义惩罚项和动态窗口路径规划算法构建语义增强型局部路径规划模型;基于四足机器人可行速度、预处理之后的多源点云融合信息和语义增强型局部路径规划模型进行可行轨迹选择,确定最佳避障规划路径。显著提升了机器人在高风险工业环境中的导航安全性与任务执行效率。
Need to check novelty before this filing date? Find Prior Art