Human Motion Analysis with Wearable Inertial Sensors

High-resolution, quantitative data obtained by a human motion capture system can be used to better understand the cause of many diseases for effective treatments. Talking about the daily care of the aging population, two issues are critical. One is to continuously track motions and position of aging people when they are at home, inside a building or in the unknown environment; the other is to monitor their health status in real time when they are in the free-living environment. Continuous monitoring of human movement in their natural living environment potentially provide more valuable feedback than these in laboratory settings. However, it has been extremely challenging to go beyond laboratory and obtain accurate measurements of human physical activity in free-living environments. Commercial motion capture systems produce excellent in-studio capture and reconstructions, but offer no comparable solution for acquisition in everyday environments. Therefore in this dissertation, a wearable human motion analysis system is developed for continuously tracking human motions, monitoring health status, positioning human location and recording the itinerary.

In this dissertation, two systems are developed for seeking aforementioned two goals: tracking human body motions and positioning a human. Firstly, an inertial-based human body motion tracking system with our developed inertial measurement unit (IMU) is introduced. By arbitrarily attaching a wearable IMU to each segment, segment motions can be measured and translated into inertial data by IMUs. A human model can be reconstructed in real time based on the inertial data by applying high efficient twists and exponential maps techniques. Secondly, for validating the feasibility of developed tracking system in the practical application, model-based quantification approaches for resting tremor and lower extremity bradykinesia in Parkinson’s disease are proposed. By estimating all involved joint angles in PD symptoms based on reconstructed human model, angle characteristics with corresponding medical ratings are employed for training a HMM classifier for quantification. Besides, a pedestrian positioning system is developed for tracking user’s itinerary and positioning in the global frame. Corresponding tests have been carried out to assess the performance of each system.