{
  "paper_id": "ecf2eeb4d5104e5885520e7958d81acf",
  "paper_title": "Activity Detection in Indoor Environments Using Multiple 2D Lidars",
  "paper_authors": [
    "Mondher Bouazizi",
    "Alejandro Lorite Mora",
    "Kevin Feghoul",
    "Tomoaki Ohtsuki"
  ],
  "publication_date": "2024-01-18",
  "publication_venue": "Sensors",
  "abstract": "In health monitoring systems for the elderly, a crucial aspect is unobtrusively and continuously monitoring their activities to detect potentially hazardous incidents such as sudden falls as soon as they occur. However, the effectiveness of current non-contact sensor-based activity detection systems is limited by obstacles present in the environment. To overcome this limitation, a straightforward yet highly efficient approach involves utilizing multiple sensors that collaborate seamlessly. This paper proposes a method that leverages 2D Light Detection and Ranging (Lidar) technology for activity detection. Multiple 2D Lidars are positioned in an indoor environment with varying obstacles such as furniture, working cohesively to create a comprehensive representation of ongoing activities. The data from these Lidars is concatenated and transformed into a more interpretable format, resembling images. A convolutional Long Short-Term Memory (LSTM) Neural Network is then used to process these generated images to classify the activities. The proposed approach achieves high accuracy in three tasks: activity detection, fall detection, and unsteady gait detection. Specifically, it attains accuracies of 96.10%, 99.13%, and 93.13% for these tasks, respectively. This demonstrates the efficacy and promise of the method in effectively monitoring and identifying potentially hazardous events for the elderly through 2D Lidars, which are non-intrusive sensing technology.",
  "keywords": [
    "activity detection",
    "human activity recognition",
    "fall detection",
    "healthcare",
    "2D Lidar",
    "deep learning",
    "machine learning"
  ],
  "paper_institutions": [
    "Faculty of Science and Technology, Keio University",
    "Graduate School of Science and Technology, Keio University",
    "UMR-S1172—Lille Neuroscience and Cognition, Centre Hospitalier Universitaire Lille, Inserm, University of Lille"
  ],
  "output_language": "zh",
  "provider": "doubao",
  "model": "ark-code-latest",
  "prompt_template": "simple",
  "extracted_at": "2026-08-16T19:46:14.109446Z",
  "template_tag": "simple",
  "summary": "### 摘要\n本文针对老年健康监测中活动检测受环境障碍物影响的问题，提出了一种利用多个2D Lidar传感器协同工作进行室内活动检测的方法。该方法将多个Lidar采集的数据在早期阶段融合，并转换为类似图像的二维表示，随后使用卷积长短期记忆（ConvLSTM）神经网络进行分类。实验在九个不同室内环境（包含家具等障碍）中进行，使用Unity模拟器和AMASS数据集生成动画，分别针对活动检测、跌倒检测和不稳定步态检测三个任务进行验证。结果显示，所提方法在三个任务上的准确率分别达到96.10%、99.13%和93.13%，显著优于单Lidar和传统方法（如[17][18]），尤其在有障碍物和Lidar靠近墙边的现实场景中表现突出。核心机制包括：基于空扫描的聚类识别用户数据点，转换为绝对坐标后投影到网格生成图像，以及采用ConvLSTM网络提取空间和时间特征。该方法具有可解释性强、抗遮挡能力好等优点，但也存在Lidar间干扰、数据丢失等局限性，未来工作将考虑多Lidar频率配置和故障容错。\n### 关键要点\n- 多Lidar协同工作：多个2D Lidar放置在房间周边，通过数据融合克服单传感器被障碍物遮挡的问题，形成更全面的活动表示。\n- 数据表示创新：将来自多个Lidar的扫描数据转换为二维图像（通过聚类、坐标转换和网格化），该表示易于人类理解，增强了模型的可信度。\n- 模型架构：采用ConvLSTM神经网络，包含两个时间分布卷积层（32和64个3x3滤波器）和MaxPooling，后接LSTM层（200单元）和两个全连接层（128和N_C个神经元）。\n- 最优设置：通过实验确定最优网格尺寸为90x60，最优Lidar高度为35 mm（接近实际设备RPLidar A3M1的安装高度）。\n- 融合策略对比：早期数据融合（将原始数据拼接成图像）准确率96.10%，远高于晚期融合（74.12%），表明早期融合在活动检测中优势明显。\n- 数据集构成：使用九个不同室内环境和九个人体模型（五男四女），训练和测试数据分别来自不同房间和不同人物，确保模型泛化能力。每个活动训练样本500个，测试样本400个。\n- 跌倒检测表现：在跌倒检测任务中，整体准确率达99.13%，混淆矩阵显示“Fall”类别的精确率和召回率分别为99.00%和99.25%。\n- 不稳定步态检测较难：整体准确率93.13%，其中“Unsteady Gait”类别的召回率相对较低（91.13%），且当Lidar放置高度超过100 mm时性能显著下降（如200 mm时准确率仅65.7%）。\n- 与传统方法比较：在与单个Lidar方法（[17][18]）的对比中，所提方法准确率高出10%以上；即使用全部Lidar数据，传统方法最高仅84.88%（Bouazizi et al. [18]），而所提方法达96.10%。\n- 局限性：实际Lidar存在测量误差（0.5-2%）和数据丢失（0-5%），且多Lidar间存在相互干扰，目前仿真未完全模拟这些因素；此外，若部分Lidar故障，性能可能大幅下降。\n- 未来方向：计划进行更多真实环境实验，解决Lidar间干扰、距离估计不准和数据丢失问题，以增强方法的鲁棒性和可靠性。\n- 应用场景：适用于养老院、居家护理等场景，通过非接触式传感器监控老年人活动，及时发现跌倒等危险事件，同时保护隐私。"
}