Real-Time Open Source Kinematic Estimation with Wearable IMUs.

Abstract

Despite the growing demand for healthcare services due to an aging population, patients may avoid traditional rehabilitation centers due to high costs, discomfort, and time constraints associated with in-person assessments. Home-based rehabilitation offers a promising alternative, but effective kinematic monitoring and assessment remain challenging, especially for real-time applications. To address this gap, we have developed real-time, full-body kinematic analysis and visualization based on 12 wearable inertial measurement units (IMUs). Full body real-time kinematics estimation (20 Hz) was evaluated during walking, running, squatting, boxing, yoga, dance, badminton, and various seated extremity exercises and IMU estimations were compared with optical motion capture to determine accuracy. Results showed that walking was the most accurate with 5.4 deg median RMSE, and the overall median RMSE was 7.2 deg for all activities. A mean of 1.0 deg RMSE against offline computations (100 Hz) was also demonstrated, with a mean latency of 44.1 ms from sensor data acquisition to kinematic output. This approach holds the potential to revolutionize rehabilitation by enabling rapid assessment and real-time biofeedback for motion performance in orthopedic and neurological conditions and could significantly enhance treatment outcomes and patient compliance.