Laboratory of Instrumentation, University of Science and Technology Houari Boumediene, BP 32 El Alia, 16111, Bab Ezzouar Algiers, Algeria.
IFSTTAR, LBMC UMR_T9406, Univ Lyon, Université Claude Bernard Lyon 1, 43 Boulevard du 11 novembre 1918, F69622, Villeurbanne, France.
Med Biol Eng Comput. 2019 Nov;57(11):2449-2460. doi: 10.1007/s11517-019-02033-7. Epub 2019 Aug 30.
A lot of attention has been paid to wearable inertial sensors regarded as an alternative solution for outdoor human motion tracking. Relevant joint angles can only be calculated from anatomical orientations, but they are negatively impacted by soft tissue artifact (STA) defined as skin motion with respect to the underlying bone; the accuracy of measured joint angle during movement is affected by the ongoing misalignment of the sensor. In this work, a new sensor-to-segment calibration using inertial measurement units is proposed. Inspired by the multiple calibration for a cluster of skin markers, it consists in performing first multiple static postures of the upper limb in all anatomical planes. The movements that affect sensor alignment are identified then alignment differences between sensors and segment frames are calculated for each posture and linearly interpolated. Experimental measurements were carried out on a mechanical model and on a subject who performed different movements of right elbow and shoulder. Multiple calibration showed significant improvement in joint angle measurement on the mechanical model as well as on human joint angle comparing to those obtained from attached sensors after technical calibration. During shoulder internal-external rotation, the maximal error value decreased more than 50% after correction. Graphical abstract Elbow flexion-extension joint angle values obtained from IMUs are well-corrected after applying multiple calibration procedure. Though shoulder internal-external rotation joint angle is more affected by soft tissue artifact, multiple calibration procedure improves the angle values obtained from IMUs.
人们已经关注到可穿戴惯性传感器，它被视为一种替代方案，可以用于户外人体运动跟踪。相关关节角度只能通过解剖方位来计算，但它们会受到软组织伪影（STA）的负面影响，定义为相对于底层骨骼的皮肤运动；在运动过程中，测量的关节角度的准确性会受到传感器持续不对准的影响。在这项工作中，提出了一种新的使用惯性测量单元的传感器到段校准方法。它受多个皮肤标记集群校准的启发，包括首先在所有解剖平面上执行上肢的多个静态姿势。然后识别影响传感器对准的运动，并为每个姿势计算传感器和段框架之间的对准差异，并进行线性内插。实验测量在机械模型和执行右肘和肩部不同运动的受试者上进行。与经过技术校准后从附着传感器获得的关节角度相比，多校准在机械模型和人体关节角度测量方面显示出了显著的改进。在肩部内收/外展时，经过校正后，最大误差值降低了 50%以上。
