一种异构传感器的地图融合构建方法

A Map Fusion Construction Method for Heterogeneous Sensors

针对单一传感器建图时还原度较低,易受环境干扰及可靠性不足等问题,提出一种任务层面融合激光雷达与深度相机的建图方法。采用一种针对走廊环境的组合式对齐算法对地图进行对齐,利用基于贝叶斯推理的方法进行信息融合。为验证融合建图效果,通过搭建多样化虚拟场景进行仿真和对真实走廊环境进行地图融合实验。结果表明,经过融合的栅格地图比单一传感器创建的栅格地图还原度更高,有效提高了地图的精度,得到了更能反映真实物理环境信息的栅格地图,为后续的路径规划提供了准确可靠的环境地图。

A task level mapping method that integrates Laser radar and depth camera is proposed to address the issues of low fidelity,susceptibility to environmental interference,and insufficient reliability when using a single sensor for mapping.Firstly,a combination based algorithm for corridor environment is adopted to align the map,and then a Bayesian inference based method is used for information fusion.To verify the effectiveness of fusion mapping,simulations are conducted by building diverse virtual scenes and map fusion experiments are carried out on real corridor environments.The results show that the fused grid map has a higher degree of restoration than the grid map created by a single sensor,effectively improving the accuracy of the map and obtaining a grid map that better reflects real physical environment information.This provides accurate and reliable environmental maps for subsequent path planning.