Angle measurement using inertial sensors
Within the scope of the research project, a method for measuring human joint angles using two inertial
sensors (humotion GmbH, Münster, Germany) was investigated. A goniometer (biovision, Wehrheim,
Germany) was used as a reference system. The objective was to keep the difference between the
measured values of both systems as small as possible. The measurements were carried out with
twelve adult test persons. First, the angle progressions of the knee joint were recorded, while the
subjects completed eleven different movement tasks (e.g. walking, climbing stairs, jumping).
The data evaluation was carried out with the WEKA Data Mining Software (University of Waikato, New Zealand), where the measured data of both systems were related to each other for eleven subjects using different algorithms. In this way, it was finally possible to calculate unknown joint angle progressions of a knee based on measured data from the inertial sensors.
On the twelfth test data set the differences of the maximum flexion and extension values between calculated angle and simultaneously measured goniometer angle were analyzed. For the extension movement, the mean values for all tasks of this data set showed an average difference of 4.5° ± 2.9°, the maximum deviation was 9.9°. Knee flexion showed an average deviation of 4.5° ± 4.4° with a maximum deviation of 13.8°. The following figure shows the results for the "climbing stairs" test task as an example. The measured curve could be reproduced with high accuracy for most of the measuring points.
The data evaluation was carried out with the WEKA Data Mining Software (University of Waikato, New Zealand), where the measured data of both systems were related to each other for eleven subjects using different algorithms. In this way, it was finally possible to calculate unknown joint angle progressions of a knee based on measured data from the inertial sensors.
On the twelfth test data set the differences of the maximum flexion and extension values between calculated angle and simultaneously measured goniometer angle were analyzed. For the extension movement, the mean values for all tasks of this data set showed an average difference of 4.5° ± 2.9°, the maximum deviation was 9.9°. Knee flexion showed an average deviation of 4.5° ± 4.4° with a maximum deviation of 13.8°. The following figure shows the results for the "climbing stairs" test task as an example. The measured curve could be reproduced with high accuracy for most of the measuring points.
In general, the highest differences occurred when the plane in which the measurement took place changed (e.g. when rotating). Movements in one plane could be calculated with high accuracy. For all test tasks involving different types of walking or running, the angle progressions could be precisely predicted.