GDTM

GDTM是由加州大学洛杉矶分校创建的一个室内地理空间跟踪数据集，包含九小时的分布式多模态传感器数据。该数据集通过多种传感器（如立体视觉相机、LiDAR相机、毫米波雷达和麦克风阵列）收集，旨在为室内环境中的自主系统设计提供支持。数据集的创建过程中，使用了网络时间协议（NTP）同步传感器网络，并设计了软件工具来控制数据收集。GDTM的应用领域包括优化多模态数据处理架构、研究模型对不利感知条件和传感器位置变化的鲁棒性等，旨在解决室内自主系统的关键定位和跟踪问题。GDTM is an indoor geospatial tracking dataset developed by the University of California, Los Angeles (UCLA), containing nine hours of distributed multimodal sensor data. Collected via a variety of sensors including stereo vision cameras, LiDAR cameras, millimeter-wave radars and microphone arrays, it is designed to support the development of autonomous systems in indoor environments. During the dataset's creation, the Network Time Protocol (NTP) was employed to synchronize the sensor network, and dedicated software tools were developed to control data collection. The application scenarios of GDTM include optimizing multimodal data processing architectures, investigating model robustness against adverse perceptual conditions and changes in sensor placement, among others, with the goal of addressing key localization and tracking challenges for indoor autonomous systems.

搜集汇总

背景与挑战

背景概述

GDTM是由加州大学洛杉矶分校创建的室内地理空间跟踪数据集，包含九小时的多模态传感器数据，使用立体视觉相机、LiDAR相机等多种传感器收集，旨在支持室内自主系统的设计，应用于优化数据处理架构和研究模型鲁棒性，以解决定位和跟踪问题。