Quantifying and predicting elevation angle error using tri-axial accelerometer during dynamic motion.
Proceedings of the 4th North American Congress on Biomechanics, August 5-9, 2008, Ann Arbor, Michigan (32nd Annual Meeting of the American Society of Biomechanics in conjunction with the 15th Conference of the Canadian Society for Biomechanics). Newark, DE: The American Society of Biomechanics and Ottawa, ON, Canada: The Canadian Society for Biomechanics, 2008 Aug; :238
INTRODUCTION: Linear accelerometers are commercially available and are commonly used in evaluation of segments' posture by means of uni-axial (Paquet et al. 2001), bi-axial (Boonstra et al. 2006) and tri-axial (Hansson et al. 2001) accelerometers. However, the main problem with linear accelerometers is that any non-gravity linear acceleration will bias the calculated elevation angles. Therefore the purpose of this study is to test and evaluate triaxial accelerometer accuracy under dynamic conditions and the ability to predict the elevation angle error. METHODS AND PROCEDURES: The Virtual Corset, (Microstrain Inc,VT, USA) is a pager sized, battery powered triaxial accelerometer with an integrated 2 Mb data logger and a sampling rate of approximately 7.6 Hz. A SW22B Wirewound precision single turn potentiometer (ETI Systems Inc, CA, USA), with a linearity tolerance of +/- 0.5%, was connected to an aluminum arm to create a pendulum. It was felt that a pendulum would result in a controlled environment that introduced high and variable levels of angular velocities and accelerations. To predict the angle error (-0-) in elevation angle, the angle between the actual resultant and gravity acceleration vectors was calculated. This equation can be represented by angular velocity ( ) and acceleration (a), radius (r), and elevation angle (B). To check the accuracy of this prediction equation the VC was mounted on the pendulum's arm at nine different radii: 0- 10cm in 2cm increments and 0-25cm in 5 cm increments. For each trial, the pendulum's arm was released from an angle of 105 degrees of elevation and data were collected from the VC and potentiometer for 15 seconds and saved. The potentiometer data were sampled at 1000Hz. These settings were repeated for each of the VC at three different positions which represent, the frontal, scapular and sagittal planes. The actual angle error and the predicted angle error were compared. To validate the use of the VC beyond the pendulum setting for in vivo measurement, data from a previous reaching study were used. The subjects performed a controlled continuous seven arm elevations and depression (Constrained) and also two unconstrained reaching movements of reaching overhead (Overhead) and reaching to a seat belt (Belt). These data were used to calculate the averaged RMS and absolute maximum predicted angle errors for each task. RESULTS: Under in-vitro dynamic conditions the calculated elevation angle error was increased as the radius increased and as the angular acceleration increased. The calculated predicted elevation angle errors from the pendulum's data were found to be similar to the VC calculated elevation angle errors. Using a radius of 10cm on the in-vivo previously collected data, constrained arm elevation had the lowest averaged RMS and maximum predicted angle errors. DISCUSSION: The farther the VC is located from the axis of rotation the higher the errors. The same was true for larger angular accelerations. The angular velocity did not have a large impact under these settings because the radial acceleration was parallel to the gravitational acceleration vector. It was also found that plane of elevation did not affect the error. Our proposed prediction equation has the ability to predict the error, which may help the investigator to make a decision on how appropriate the VC is for measuring exposure in a specific job environment. SUMMARY: The VC can be used to reconstruct elevation angles. In order to improve data collection qualities we offer the following recommendations: 1. Locate the VC as close as possible to the joint center of rotation. 2. Estimate the maximum and average angular velocity and acceleration of the task. 3. Determine the typical and maximal range of humeral elevation angle. 4. Use the prediction equation to determine whether the expected errors are within acceptable tolerances for the given experiment.
Sampling; Sampling-equipment; Measurement-equipment; Acceleration; Posture; Analytical-processes
Proceedings of the 4th North American Congress on Biomechanics, August 5-9, 2008, Ann Arbor, Michigan (32nd Annual Meeting of the American Society of Biomechanics in conjunction with the 15th Conference of the Canadian Society for Biomechanics)
University of Oregon