arXiv 2505.074462025-07-09cs.RO

TPT-Bench：用于目标人物跟踪基准测试的大规模、长期、机器人自我中心数据集

TPT-Bench: A Large-Scale, Long-Term and Robot-Egocentric Dataset for Benchmarking Target Person Tracking

Robotics and Computer Vision Laboratory, Southern University of Science and Technology, Shenzhen, China（机器人与计算机视觉实验室，南方科技大学，深圳，中国）；Human-Robot Interfaces and Interaction Laboratory, Italian Institute of Technology, Geona, Italy（人机交互实验室，意大利理工学院，Geona，意大利）；Robotics Perception and Intelligence Laboratory, Southern University of Science and Technology, Shenzhen, China（机器人感知与智能实验室，南方科技大学，深圳，中国）；Chen Kar-Shun Robotics Institute, Hong Kong University of Science and Technology, Hong Kong SAR, China（陈卡尔顺机器人研究所，香港科技大学，香港特别行政区，中国）

Hanjing Ye, Yu Zhan, Weixi Situ, Guangcheng Chen, Jingwen Yu, Ziqi Zhao, Kuanqi Cai, Arash Ajoudani, Hong Zhang

备注：: Under review. web: https://medlartea.github.io/tpt-bench/

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

本文提出TPT-Bench，一个大规模、长期、机器人自我中心的数据集，用于在拥挤非结构化环境中基准测试目标人物跟踪，并分析现有SOTA方法的局限性。

摘要：

从机器人自我中心视角跟踪目标人物对于开发能够在人机交互（HRI）和具身AI中提供持续个性化协助或协作的自主机器人至关重要。然而，现有的大多数目标人物跟踪（TPT）基准仅限于受控实验室环境，这些环境具有少量干扰、干净背景和短期遮挡。在本文中，我们介绍了一个专为拥挤和非结构化环境中的TPT设计的大规模数据集，并通过机器人-人物跟随任务进行演示。该数据集由一名人类推着配备传感器的推车跟随目标人物收集，捕捉类似人类的跟随行为，并强调长期跟踪挑战，包括频繁遮挡以及从众多行人中重新识别的需求。它包含多模态数据流，包括里程计、3D LiDAR、IMU、全景图像和RGB-D图像，以及跨48个序列（室内和室外）的目标人物详尽标注的2D边界框。利用该数据集和视觉标注，我们使用现有SOTA TPT方法进行了广泛实验，对其局限性进行了深入分析，并提出了未来研究方向。

英文摘要：

Tracking a target person from robot-egocentric views is crucial for developing autonomous robots that provide continuous personalized assistance or collaboration in Human-Robot Interaction (HRI) and Embodied AI. However, most existing target person tracking (TPT) benchmarks are limited to controlled laboratory environments with few distractions, clean backgrounds, and short-term occlusions. In this paper, we introduce a large-scale dataset designed for TPT in crowded and unstructured environments, demonstrated through a robot-person following task. The dataset is collected by a human pushing a sensor-equipped cart while following a target person, capturing human-like following behavior and emphasizing long-term tracking challenges, including frequent occlusions and the need for re-identification from numerous pedestrians. It includes multi-modal data streams, including odometry, 3D LiDAR, IMU, panoramic images, and RGB-D images, along with exhaustively annotated 2D bounding boxes of the target person across 48 sequences, both indoors and outdoors. Using this dataset and visual annotations, we perform extensive experiments with existing SOTA TPT methods, offering a thorough analysis of their limitations and suggesting future research directions.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。