IntelliMan_WP5_Grasping, Manipulation and Arm-Hand Coordination_T5.1_Data fusion and sensing technology_IEEEICECS2024_Data

This dataset contains multisensory time-series recordings acquired using a wearable glove designed for object recognition during grasping interactions. The glove integrates commercially available sensing components, including five force-sensing resistors (FSR402) positioned on the fingertips and five 6-axis inertial measurement units (MPU6050) mounted on the back of the distal phalanges. The sensing system provides a total of 35 acquisition channels, comprising five force channels and 30 inertial channels (three accelerometer axes and three gyroscope axes per IMU). The dataset was collected from five participants interacting with 28 everyday objects. Each participant performed 10 grasp–release trials per object, resulting in a total of 1,400 observations, with 50 trials per object class. During each trial, participants grasped an object for approximately one second before releasing it. Sensor signals were acquired at a sampling frequency of 200 Hz, resulting in 200 temporal samples per channel. Each observation is represented as a two-dimensional array of dimensions 200 × 35, where the first dimension represents the temporal evolution of the signals and the second corresponds to the sensing channels. The dataset enables the investigation of multisensory fusion as well as the individual contributions of force and inertial measurements to object recognition. The dataset was used to investigate embedded machine learning approaches for real-time object recognition, employing a lightweight one-dimensional convolutional neural network (1D-CNN) operating directly on raw multisensory signals. The proposed classification framework was deployed on an STM32F401-RE microcontroller to evaluate recognition accuracy, inference latency, and energy consumption. The dataset is associated with the following peer-reviewed conference publication, which provides a detailed description of the wearable sensing system, experimental protocol, data acquisition methodology, and embedded machine learning framework: M. Yaacoub, H. Al Haj Ali, C. Gianoglio, M. Valle, and A. Ibrahim, "Multisensory Wearable Glove for Object Recognition Based Embedded Machine Learning," 2024 31st IEEE International Conference on Electronics, Circuits and Systems (ICECS), 2024. DOI: https://doi.org/10.1109/ICECS61496.2024.10848876 Users of this dataset are kindly requested to cite the above publication when using the data in their research.