室内商场视觉导航避障数据

面向基于视觉的机器人环境建模与定位导航基础研究，为突破复杂环境下机器人视觉导航避障技术瓶颈，基于双目RGB彩色相机，在北京奥特莱斯（场地占地面积约14000平方米，存在动态目标，光线条件较为复杂，以自然光和人工光照混合条件为主）进行数据采集。该数据集包含1426张彩色影像（分辨率为640×360像素,采样频率5Hz），数据量为435M，主要构建复杂场景下的商场视觉导航避障据集。This dataset is developed for fundamental research on vision-based robotic environment modeling, localization and navigation. To break through the technical bottlenecks of robotic visual navigation and obstacle avoidance in complex environments, data collection was carried out at Beijing Outlets, a site covering an area of approximately 14,000 square meters with dynamic targets and complex illumination conditions primarily consisting of a mix of natural and artificial lighting, using a binocular RGB color camera. The dataset contains 1,426 color images with a resolution of 640×360 pixels and a sampling frequency of 5 Hz, with a total data volume of 435 MB. It is primarily constructed as a visual navigation and obstacle avoidance dataset for shopping mall scenarios in complex environments.

搜集汇总

背景与挑战

背景概述

该数据集面向机器人视觉环境建模与定位导航研究，通过双目RGB相机在北京奥特莱斯（面积约14000平方米，包含动态目标和复杂光照条件）采集，包含1426张彩色影像（分辨率640×360像素，采样频率5Hz），数据量435MB，旨在构建商场复杂场景下的视觉导航避障数据集。