Skip to main content
Utah's Foremost Platform for Undergraduate Research Presentation
2018 Abstracts

Comparison of two common methods used to calibrate motion capture systems

Ryan Clark; Taylor Dickinson; Johnfredy Loaiza; Kari Beardsley; Dan Geiger, Brigham Young University

Electromagnetic (EM) motion tracking systems have been tested extensively and found suitable for many research and clinical applications. Such systems consist of a transmitter and multiple sensors that provide the position and orientation of each sensor relative to the transmitter. While it is easy to attach these sensors to limb segments and take measurements, extracting accurate and useful data (i.e. joint angles) from the sensors can be challenging. To do so a large and relatively complex set of steps must be followed: proper definition of Body Coordinate Systems (BCS) and Joint Coordinate Systems (JCS), proper sensor placement, the calculations governing the inverse kinematics, and the choice between different methods used to calibrate the Sensor Coordinate Systems (SCS) to the BCS. This study is a comparative analysis to understand how similar two commonly used calibration methods are: Landmark (LM) and Postural (P) calibrations. These two methods have intrinsic differences in BCS definitions. LM calibration uses the locations of skeletal landmarks relative to the SCS to define the BCS. It is commonly used by biomechanists in joint-level and cadaveric studies. P calibration uses superficial alignment of the limb segments in a known configuration to determine the BCS. P is commonly used by clinicians who study gross limb movement of subjects in-vivo. We collected EM motion tracking data from 12 healthy subjects as they performed a sequence of movements throughout their range of motion. EM sensor orientations were tracked and recorded. Inverse kinematics were performed on the same sensor data using both LM and P calibration methods. The joint angles resultant from LM and from P were then compared. The results indicated that the two methods differ in two major ways. First, there is an initial offset that is the result of innate differences in BCS definitions of LM and P. This offset is large enough in magnitude to make a significant impact on studies requiring accurate joint angle measurements. Second, the difference in joint angle values between the two calibration methods was not equal to the initial offset. Although there was high correlation in joint angle values between the calibration methods, the difference varied during the movements. Therefore, the two calibration methods are not equivalent, even for applications that focus only on relative motion (e.g. tremor). Caution should be used when choosing a method of calibration for upper-limb studies.