Open source platform for collaborative construction of wearable sensor datasets for human motion analysis and an...

Area of Science:

• Biomechanics and Human Motion Analysis
• Wearable Sensor Technology
• Data Science and Machine Learning

Background:

• Advancements in human motion modeling rely heavily on well-designed, publicly accessible datasets for validation.
• Current limitations in dataset size and scope hinder comprehensive community validation and acceptance.
• Collaborative efforts, similar to the open-source software community, are needed to build robust datasets.

Purpose of the Study:

• To introduce an affordable, open-source hardware and software platform for collaborative data acquisition.
• To enable multiple research groups to cooperate in constructing shared human motion datasets.
• To facilitate the expansion and enrichment of human motion datasets through shared resources.

Main Methods:

• Development of an integrated open-source platform utilizing wearable inertial sensors.
• Implementation of common software to ensure data acquisition compatibility among collaborators.
• Experimental validation of system throughput for multi-sensor data acquisition at high frequencies.

Main Results:

• Demonstrated feasibility of acquiring data from up to 6 inertial sensors.
• Achieved a sampling frequency of no less than 118Hz, ensuring high-fidelity data capture.
• Provided a proof-of-concept dataset with sampled data from 12 subjects for gait analysis.

Conclusions:

• The proposed platform offers a viable solution for collaborative dataset construction in human motion research.
• The system's affordability and open-source nature promote wider community participation and data sharing.
• The generated dataset serves as a foundation for further research in gait analysis and human motion modeling.