NCLT

该数据集是由密歇根大学北校区在不同时间段和不同季节操作一辆赛格威机器人收集的实证数据组成，数据包括了来自前向摄像头的图像和用于地面真实重建的GPS数据。图像中涵盖了各种不同的景象，如积雪和落叶，并且该数据集已经预处理过了，以实现视角轨迹的离散化。该数据集的规模包括五个测试会话，共计5000个片段。其任务是对视觉地点识别（Vpr）。

This dataset consists of empirical data collected by operating a Segway robot on the North Campus of the University of Michigan across different time periods and seasons. The data includes images from the forward-facing camera and GPS data for ground-truth reconstruction. The images cover various scenarios such as snowy environments and leaf-covered areas, and the dataset has been preprocessed to enable the discretization of the viewpoint trajectory. It includes five test sessions totaling 5000 segments, and its targeted task is visual place recognition (VPR).

搜集汇总

背景与挑战

背景概述

NCLT数据集是一个长期、多传感器的机器人数据集，包含激光雷达、视觉、IMU和GPS等多种数据，覆盖15个月的时间和147.4公里的距离，适用于SLAM和机器人导航研究。