Extensive work has been done to reduce discrepancies between kinematics obtained from IMUs with those obtained from optoelectronic systems. These studies dealt with sensor-to-segment calibration (Pacher et al., 2020), the use of Multibody Kinematics Optimization (MKO) (Kok et al., 2014, Lin and Kulić, 2012, Weygers et al., 2020, Pacher et al., 2023), and improving IMU data-fusion algorithms (Nazarahari and Rouhani, 2021). However, these studies seem to reach a “barrier” since it is noticeable that the differences between joint kinematics obtained from IMUs and those obtained from optoelectronic reference remain noteworthy, especially on the frontal and transverse planes (Mcgrath and Stirling, 2022, Pacher et al., 2023). To explain these results, it has been suggested that IMUs could be particularly sensitive to the Soft Tissue Artefact (STA) (Kamstra et al., 2022, Pacher et al., 2023).

STA denotes skin surface deformation (muscle contractions, skin sliding and stretching), inertial and dynamical effects (wobbling masses) which cause differences on the kinematics of segment skin surfaces and underlying bones (Barré et al., 2015, Stagni et al., 2005). With markers, these STA effects could be seen as marker-cluster deformation (i.e. relative movement of one marker with respect to the others) and marker-cluster rigid transformation (i.e. rotation and translation of all the markers of a segment, also depicted as in-unison or rigid component), the last one being the major (Barré et al., 2015, Duprey et al., 2017). With IMUs, when one IMU per segment is used, these STA effects cannot be separated into deformation and rigid components.

To try to reduce STA effects with optoelectronic data, kinematics optimization methods is often employed. The first method called single-body (or segmental) kinematics optimization (SKO) uses markers redundancy and a rigid-body hypothesis to reduce STA at a segment level (Chèze et al., 1995), by considering marker cluster deformation. The other one, the MKO, is based on kinematic constraints (Begon et al., 2018), and minimizes difference between a model-derived (i.e., kinematic chain of rigid segments articulated by joints) and the measured markers position.

Depending on how the joint constraints are set up and the level of subject-specific joint modelling, MKO exhibits mitigate (Andersen et al., 2010, Richard et al., 2017) to encouraging results (Charbonnier et al., 2017, Clément et al., 2015) to reduce STA effects around one or more segment axes.

Recently, different authors proposed MKO based on IMU data and the OpenSense plugin from OpenSim (Al Borno et al., 2022, Pacher et al., 2023). In this case, the difference to be minimized is between model-derived and measured IMUs orientations. However, only one IMU per segment was used. Even though the studies showed that this approach could be used to prevent kinematic drift (Koning et al., 2015, Luinge et al., 2007, Zhou and Hu, 2010), the question of using IMU redundancy to try to mitigate STA deformation, as for marker-based approaches, remains unanswered.

Accordingly, the main goal of this preliminary study was to test the potential of IMU redundancy when using MKO to improve lower-body kinematics obtained from IMUs.