{
  "paper_id": "16d9cc4ec6b471c57a319230ab7ec52c",
  "paper_title": "Online Learning for Human Classification in 3D LiDAR-based Tracking",
  "paper_authors": [
    "Zhi Yan",
    "Tom Duckett",
    "Nicola Bellotto"
  ],
  "publication_date": "",
  "publication_venue": "2017 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)",
  "abstract": "Human detection and tracking are essential aspects to be considered in service robotics, as the robot often shares its workspace and interacts closely with humans. This paper presents an online learning framework for human classification in 3D LiDAR scans, taking advantage of robust multi-target tracking to avoid the need for data annotation by a human expert. The system learns iteratively by retraining a classifier online with the samples collected by the robot over time. A novel aspect of our approach is that errors in training data can be corrected using the information provided by the 3D LiDAR-based tracking. In order to do this, an efficient 3D cluster detector of potential human targets has been implemented. We evaluate the framework using a new 3D LiDAR dataset of people moving in a large indoor public space, which is made available to the research community. The experiments analyse the real-time performance of the cluster detector and show that our online learned human classifier matches and in some cases outperforms its offline version.",
  "keywords": [
    "online learning",
    "human classification",
    "3D LiDAR",
    "multi-target tracking",
    "P-N experts",
    "service robotics"
  ],
  "paper_institutions": [
    "Lincoln Centre for Autonomous Systems, University of Lincoln, UK"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:47:34.614518Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本文面向服务机器人中的人体检测与跟踪问题，提出了一种基于3D LiDAR扫描的在线学习框架。该框架利用鲁棒的多目标跟踪系统，避免了人工标注训练数据的负担，通过机器人随时间采集的样本迭代地重训练分类器。其新颖之处在于利用3D LiDAR跟踪提供的信息纠正训练数据中的错误，为此实现了一个高效的3D点簇检测器。作者在大型室内公共空间采集了新的3D LiDAR数据集（包含人类群体、儿童、携带手推车的人等挑战），并公开供研究社区使用。实验分析了簇检测器的实时性能，并表明在线学习得到的人体分类器在部分指标上匹配甚至超越离线训练的分类器。框架基于SVM分类器、UKF跟踪器和P-N专家样本生成器，采用批增量学习，通过位置和速度约束识别误检和漏检，在真实机器人平台上验证了其有效性和稳定性。未来工作将扩展至其他移动目标和其他传感器，并改进遮挡和群体场景下的检测性能。\n\n### 关键要点\n- 提出了一个结合3D LiDAR点簇检测、多目标跟踪、人体分类器和样本生成器的在线学习框架，通过P-N专家纠正分类错误。\n- 设计了基于自适应距离阈值的3D点簇检测算法，阈值随扫描距离动态调整（d*=2r·tan(Θ/2)），并利用嵌套环状区域提高计算效率。\n- 使用UKF和最近邻数据关联进行多目标跟踪，采用2D极坐标观测模型，更好地匹配LiDAR传感器特性。\n- 分类器采用SVM，使用6类特征共61维（点数、距离、协方差、惯性张量、切片特征和反射强度分布），经过实验验证适用于实时应用。\n- 训练采用批增量方式，每收集一定数量的正负样本后重新训练，避免了逐帧更新的开销。\n- P专家基于轨迹的移动距离和速度（0至1.4 m/s）识别漏检，N专家利用静止物体低位置方差识别误检，并设置不确定性阈值过滤不可靠样本。\n- 实验数据集包含28,002帧，来自Velodyne VLP-16，室内场景，含5,492帧手工标注，标注了6,140个行人标签。\n- 簇检测器在全程实时运行，平均频率随距离下降但始终满足≥5 Hz的跟踪要求，最远检测距离约25 m。\n- P-N专家性能分析表明P专家精度97.1%，N专家精度61.6%，总错误数满足稳定性条件。\n- 分类对比显示在线学习分类器的最终性能显著优于初始分类器，并匹配或超越离线训练分类器，尤其受益于远距离样本的自动获取。\n- 框架已实现为模块化ROS包，代码和数据集已公开，便于复现和扩展至其他移动目标或传感器。",
  "available_templates": [
    "deep_read",
    "simple"
  ]
}