Inertial Sensor Motion Tracking

Abstract

The recent advancements in inertial sensors technology and its promising results in motion tracking, catch the expert’s eyes to these new horizons in sports engineering. In baseball, which is the interest of this study, almost 90% of the pitchers got injured once a year due to wrong training and pitching techniques. Screening and real-time feedback to players would help them to improve their training procedures and safely increase their pitching performance.
In the last decades many researches have been carried out on employing this new tool in field measurements, instead of the common marker-based motion tracking with their complexity for the in field measurements. Although it is promising that inertial sensors are the future of motion tracking systems in this area, there are still many technical issues like, IMU sensors measurement limit, drift and bias. Besides in human motion tracking which involves multiple IMU sensors, coinciding the sensors and defining a global coordinate system requires substantial concerns.
This study focuses on developing a valid motion tracking method for the baseball pitchers, having a marker-based motion capture measurement as the reference. In order to be able to do this first of all the two systems was needed to be synchronized and at the same time be able to record the same motion. Secondly, the measurements should be defined in the same coordinate system. For this purpose, a simple functional calibration method has been developed and applied on both systems. This method is validated against a previous method (Seel, Schauer et al. 2012). Finally, The kinematic results are estimated at joint and segment’s angles, velocities and accelerations levels. The joint and segment’s angles computed by IMU sensors are validated based on marker-based measurements. The sensitivity of IMU-based measurements in estimating the angular velocity and acceleration of movements with different rate of movements (slow vs. fast) is investigated. It has been observed that for baseball pitching applications, IUM sensors with less mass and wider range of measurements are required.
In order to compare the dynamics of the human body, a scalable anthropometric model from the literature is used to define the mass and inertia properties of the segments. An inverse dynamics method is used to compute the kinetics energy and finally the power flow in the segments and joints. Again, all these results from the IMU measurements are compared with the Marker-based method. The advantages and disadvantages of the IMU according to these results are discussed to establish a practical protocol for future measurements and data analysis. One of the major issues in the dynamic analysis is that for translating the velocity from IMU to human body, the measurement protocol needs to provide a known starting and ending velocity. This is done by starting and ending the measurements from a standing position.
The method of this project can be used in baseball pitching motion tracking using the suggested protocol improvements and more advanced IMU sensors.