{
  "paper_id": "2d937734dfe08aee458a3285ad3af94e",
  "paper_title": "IndoorBEV: Joint Detection and Footprint Completion of Objects via Mask-based Prediction in Indoor Scenarios for Bird's-Eye View Perception",
  "paper_authors": [
    "Haichuan Li",
    "Changda Tian",
    "Panos Trahanias",
    "Tomi Westerlund"
  ],
  "publication_date": "",
  "publication_venue": "unknown",
  "abstract": "Detecting diverse objects within complex indoor 3D point clouds presents significant challenges for robotic perception, particularly with varied object shapes, clutter, and the co-existence of static and dynamic elements where traditional bounding box methods falter. To address these limitations, we propose IndoorBEV, a novel mask-based Bird's-Eye View (BEV) method for indoor mobile robots. In a BEV method, a 3D scene is projected into a 2D BEV grid which handles naturally occlusions and provides a consistent top-down view aiding to distinguish static obstacles from dynamic agents. The obtained 2D BEV results is directly usable to downstream robotic tasks like navigation, motion prediction, and planning. Our architecture utilizes an axis compact encoder and a window-based backbone to extract rich spatial features from this BEV map. A query-based decoder head then employs learned object queries to concurrently predict object classes and instance masks in the BEV space. This mask-centric formulation effectively captures the footprint of both static and dynamic objects regardless of their shape, offering a robust alternative to bounding box regression. We demonstrate the effectiveness of IndoorBEV on a custom indoor dataset featuring diverse object classes including static objects and dynamic elements like robots and miscellaneous items, showcasing its potential for robust indoor scene understanding.\n",
  "keywords": [
    "Bird's-Eye View",
    "Point Cloud",
    "Robotic Perception",
    "Indoor Scene Understanding"
  ],
  "paper_institutions": [],
  "output_language": "zh",
  "provider": "doubao",
  "model": "deepseek-v4-flash",
  "prompt_template": "simple",
  "extracted_at": "2026-08-18T10:30:17.248515Z",
  "template_tag": "simple",
  "summary": "### 摘要\n\n室内环境中的三维点云感知面临物体形状多样、场景杂乱、静态与动态目标共存等挑战，传统边界框回归难以准确描述物体足迹。已有BEV方法多面向室外自动驾驶，室内移动机器人场景缺乏专门研究。为此，本文提出IndoorBEV，一种基于掩码预测的鸟瞰图感知框架：通过轴融合编码器AF-PCCN将原始激光雷达点云聚类并投影为BEV特征图，使用窗口化SwinTransformer骨干提取空间上下文，再由基于查询的SimpleMaskHead解码器联合预测物体类别、3D位姿属性和实例足迹掩码，其中掩码由M_j = m_j · F_proj生成。训练和评估使用自建混合数据集，包含MuJoCo+Taichi仿真数据和Livox MID-360在办公室、实验室、走廊采集的真实数据，约4.5k训练帧、1k验证帧、1k测试帧；实现中聚类大小0.02m，BEV网格500×500，30个查询，9个类别，采用DETR式匈牙利匹配和分类、Dice、Focal、L1回归联合损失。在真实测试集上达到78.4% AP@0.25、63.7% AP@0.50、67.2% mIoU和64.5% PQ，验证了掩码式检测与分割的有效性。未来工作包括引入时间信息增强动态目标跟踪、与视觉数据多模态融合，以及面向计算受限硬件的模型压缩和优化。该方法面向室内移动机器人的实时导航、碰撞避免和路径规划等真实场景，但受限于点云稀疏、真实数据多样性不足和移动平台算力约束。\n\n### 关键要点\n\n- IndoorBEV提出基于掩码的鸟瞰图感知框架，直接从原始激光雷达点云联合预测物体类别、3D位姿和足迹掩码，避免边界框回归。\n- 方法包含轴融合编码器AF-PCCN、窗口化SwinTransformer骨干和基于查询的SimpleMaskHead解码器，将点云聚类后投影到BEV网格。\n- 掩码预测头通过M_j = m_j · F_proj将查询掩码嵌入与投影BEV特征相乘生成实例掩码，可适应任意物体形状。\n- 使用MuJoCo和Taichi构建高保真激光雷达仿真器，支持超过10万条射线的实时光线追踪，并生成程序化室内场景。\n- 混合数据集包含仿真数据和真实数据，真实数据由搭载Livox MID-360的移动机器人在办公室、实验室和走廊采集，约4.5k训练、1k验证、1k测试。\n- 训练采用DETR风格集合预测和匈牙利匹配，损失包含分类、Dice、Focal掩码损失以及尺寸、位置、偏航角的L1回归损失。\n- 实现细节包括聚类大小0.02m、感知范围[-5,5]×[-5,5]×[-3,3]m、BEV网格500×500、30个查询、9个类别，使用AdamW优化器训练12k迭代。\n- 在真实测试集上达到78.4% AP@0.25、63.7% AP@0.50、67.2% mIoU和64.5% PQ，验证了掩码式检测与分割的有效性。\n- 定性结果显示IndoorBEV能在杂乱和遮挡场景中准确检测静态家具和动态机器人，并比边界框更忠实描绘物体足迹。\n- 局限性包括MID-360点云稀疏、真实环境多样性有限、移动平台计算约束；未来工作包括时间信息融合、多模态融合和模型压缩部署。"
}