发表机构
Universidade Tecnológica Federal do Paraná; University of Modena and Reggio Emilia; Istituto Italiano di Tecnologia（巴拉那联邦理工大学; 摩德纳-雷焦艾米利亚大学; 意大利技术研究院）
机构由 AI 辅助整理，请以论文原文为准。AI 中文总结
该研究针对基于EMG-IMU的HRI活动识别的跨主体挑战，构建了MAGIC-HRI数据集，发现存在较大泛化差距，证实注入少量新用户样本的个性化适配可显著提升识别性能，为HRI稳健部署提供支撑。
AI 中文摘要
本文研究基于可穿戴设备的人类活动与手势识别，以支持人机交互（HRI）中的物体交接及类装配场景。使用Myo臂带采集肌电（EMG）与惯性测量单元（IMU）信号，构建了名为MAGIC-HRI（多模态活动、手势与意图采集）的新型数据集，包含53类运动的丰富分类，涵盖巴西手语（LIBRAS）数字（0-9）、手势、物体/工具交接动作（拾取/给予/握持）、工具操作任务及通用装配/静止动作，由11名参与者采集，每类10个样本（每名参与者530个样本）。通过检测EMG能量包络的肌肉激活对信号进行分割，再用滑动窗口处理；提取时域与频域特征。通过交叉验证网格搜索调优多个经典分类器，其中随机森林（Random Forest）为最优基线模型。留一主体（Leave-One-Subject-Out, LOSO）协议显示存在较大泛化差距，表明具有显著的主体依赖性。个性化适配实验表明，注入新用户的少量样本可显著提升识别性能。总体而言，本研究贡献了面向HRI的广泛多模态数据集、强调泛化的严格评估，以及实用证据，表明个性化对于实际HRI的稳健部署可能是必要的。
英文摘要
This paper investigates wearable-based recognition of human activities and gestures to support Human-Robot Interaction (HRI) in object-handover and assembly-like scenarios. Electromyography (EMG) and Inertial Measurement Unit (IMU) signals were collected using a Myo armband, culminating in a novel dataset introduced as MAGIC-HRI (Multimodal Activity, Gesture and Intention Collection) with a large taxonomy of 53 movement classes, including Brazilian Sign Language (LIBRAS) numbers (0-9), hand gestures, object/tool handover actions (pick up/give/hold), tool-manipulation tasks, and generic assembly/idle motions, collected from 11 participants with 10 samples per class (530 samples per participant). Signals are segmented by detecting muscle activation via an EMG energy envelope, then processed using sliding windows; time- and frequency-domain features are extracted. Multiple classical classifiers are tuned via cross-validated grid search, with Random Forest as the strongest baseline. A Leave-One-Subject-Out (LOSO) protocol reveals a large generalization gap, indicating substantial subject dependence. A personalized adaptation experiment suggests that injecting a small number of samples from a new user can markedly improve recognition. Overall, the study contributes a broad, HRI-driven multimodal dataset, a rigorous evaluation emphasizing generalization, and practical evidence that personalization is likely required for robust deployment in practical HRI.
CommentsData collection was approved by the Federal University of Technology-Paraná Ethics Committee (CAAE 91430125.0.0000.0177). The MAGIC-HRI (Multimodal Activity, Gesture, and Intention Collection for HRI) dataset is available at [https://github.com/ruancarminati/MAGIC-HRI-V01.git](https://github.com/ruancarminati/MAGIC-HRI-V01.git). This paper will be presented at IEEE RO-MAN 2026
