Crowdsourcing vibration data stemming from different transportation usages

Crowdsourcing vibration data stemming from different activities and transportation usages (by trains, by buses, by bicycles by walking). We present a comprehensive dataset that provides the pattern of five activities walking, cycling, taking a train, a bus or a taxi. The measurements are carried out by embedded sensor accelerometer in smartphones. The dataset offers dynamic responses of subjects carrying smartphones in varied styles as they performing the five activities through vibrations acquired by accelerometers. The dataset contains corresponding time stamps and vibrations in three directions longitudinal, horizontal, and vertical stored in an Excel Macro-enabled Workbook (xlsm) format can be used to train an AI model in a smartphone which has potentials to collect people’s vibration data and decides what movement is being conducted. Besides, with more data are received, the database can be updated and it can be fed to train the model with a larger dataset. The prevalent of the smartphone opens the door of crowdsensing which leads to the pattern of people talking public transports can be understood. Furthermore, the time consumed in each activity is available in the dataset. Therefore, with a better understanding of people using public transports, the service and schedule can be planned perceptively. Activities to obtain the dataset are jointly funded by H2020 and Hitachi Europe.本数据集为源自多种日常活动与交通出行场景的众包振动数据，覆盖步行、骑行、乘坐火车、公交及出租车五类活动。我们构建了一套完整的数据集，可呈现上述五类活动对应的振动模式。数据采集通过智能手机内置的加速度传感器（accelerometer）完成，记录了受试者以不同握持方式携带智能手机参与活动时产生的动态振动响应。数据集包含对应时间戳与纵向、横向、竖向三个方向的振动数据，以启用宏的Excel工作簿（Excel Macro-enabled Workbook，格式为xlsm）存储，可用于训练智能手机端的人工智能模型；该模型可通过采集用户振动数据识别当前进行的运动类型。此外，随着新增数据的接入，该数据库可不断更新，进而支持使用更大规模数据集对模型进行训练。智能手机的普及推动了众包感知技术的发展，使得我们能够深入解析民众的公共交通出行模式。同时，数据集还提供了每类活动的耗时信息。因此，通过更精准地掌握民众的公共交通使用行为，可实现公共交通服务与运营时刻表的精细化规划。本数据集的采集工作由欧盟地平线2020计划（H2020）与日立欧洲公司（Hitachi Europe）联合资助。