IMU-Motion: An IMU-Based Human Motion Dataset

Inertial Measurement Units (IMUs) provide a practical means for capturing human movement across diverse environments and activities. We present a long-duration, multi-IMU dataset of human motion collected using 17 Xsens IMUs sampled at 60 Hz, with negligible packet loss (at most two missing packets in a 13-min recording). The dataset includes multiple recording sessions from the same subject performing the same track, enabling within-subject and session-to-session comparisons, as well as recordings from three different subjects. The data, covers a diverse set of activities and conditions, including sitting, walking, turning, walking on elevated surfaces, ascending and descending stairs in both indoor and outdoor environments and elevator ascent. In addition to the 17 body-mounted IMUs, a reference sensor carried continuously by the subject, on their palm, was used to identify and trim activity segments. We provide raw and synchronized timestamps corroborated using the reference sensor (Sensor logger android application \cite{}) data, sensor placement and recording details, initial data-processing code, and videos documenting the data-collection procedure. The dataset is intended to support several research application, including human activity recognition, gait analysis, physical Capacity and function tests and Pedestrian Dead Reckoning (PDR) which are elaborated in this paper.