Validation of estimated orientation from inertial measurement units with an optical tracking system

Inertial measurement units are gaining importance in the field of gait analysis and human motion tracking. To enable studying gait kinematics in incomplete spinal cord injured (iSCI) patients, wearable sensors are found to be advantageous compared to optical motion capture systems (OMCs) which are the existing gold standard for gait analysis. OMCs involve the use of large hardware set ups which usually confine motion tracking to a laboratory environment. In order to reduce dependency on room-based hardware and therefore enable studying walking patterns of daily life activities in patients, inertial measurement systems (IMUs) seem to be an appropriate alternative. However, to understand the validity of results obtained by use of IMU systems, results need to be compared to those obtained from OMCs. There are various algorithms that are studied to analyse the kinematic gait parameters from IMUs. This thesis aims to validate an algorithm that do not use the magnetometer data and considers kinematic constraints in lower limb joints to obtain solutions more comparable to the OMC. This algorithm has been previously tested on mechanical hinge joints and proven to have a sensor-tosegment calibration accuracy around 2_ compared to the angular deviation from the ground truth of the joint axis [1]. In this thesis, a total of 15 individuals with no neurological disorders (control group) and 5 iSCI patients were recruited. Their walking patterns were recorded using IMUs and OMC based on a full body set up while the participants walked on a treadmill as well as on overground (straight walking, walking with turn, walking in the shape of ’8’). For the validation study, a set of sixteen 3D-printed holder that can hold one IMU and 3 optical markers per holder for each segment were used. The lower limb joint angles calculated from the IMU data were compared with the plug-in gait model of OMC. Another set of validation is made with the optical markers in the 3D-printed holder from which the orientation of the segments and the joint angles are calculated as well. A database is also built with the 15 control group participants and 5 iSCI patients as a full body set up of OMC and IMU data. This database is aimed to be provided open source to help researchers test their IMU algorithms without the need to make new measurements.