基于无人机点云地图的地面机器人重定位方法
Ground Robot Relocation Method Based on UAV Point Cloud Map
黄宏智 1颜凯 2刘昌锋 2王建文 2罗斌1
作者信息
- 1. 武汉大学 测绘遥感信息工程国家重点实验室,湖北 武汉 430070
- 2. 智能汽车安全技术全国重点实验室,重庆 401133
- 折叠
摘要
针对无GNSS(global navigation satellite system)环境下空地协同系统中地面机器人重定位难题及其精度不足的问题,提出了一种基于三维点云地图的由粗到精的重定位算法.通过索引滤波消除高空和地面无效点云的影响,并在提取点云全局特征后引入截断最小二乘估计进行粗定位,采用体素法ICP(iterative closest point)精优化,以获得更精确的定位结果.构建了一种基于空中全局地图的地面机器人定位与自主移动框架,并通过仿真平台实验证明了该框架的可行性,验证了地面机器人重定位算法的实时性和准确性.
Abstract
In response to the challenge of relocalization in air-ground collaborative systems without the support of Global Navigation Satellite System(GNSS),and the associated issues of insufficient accuracy,a coarse-to-fine relocalization algorithm based on a three-dimensional point cloud map is proposed.The algorithm eliminates the influence of invalid point clouds from the sky and ground through index filtering,performs coarse localization by extracting global features from the point cloud and applying truncated least squares estimation,and then employs voxel-based iterative closest point(ICP)for precise optimization to obtain the more accurate localization results.A ground robot localization and autonomous navigation framework is constructed based on an aerial global map and the feasibility of the framework is validated through experiments on a simulation platform,while the real-time and accuracy of the ground robot relocalization algorithm is verified.
关键词
地面机器人/重定位/无人机/全局点云地图/自主移动Key words
ground robot/relocalization/UAV/global point cloud map/autonomous navigation引用本文复制引用
基金项目
智能汽车安全技术全国重点实验室开放课题(CSTB2022TIAD-DEX0013)
出版年
2024
NETL
NSTL
万方数据