Model Based Determination of Gait Parameters Using IMU Sensor Data

Abstract

The paper deals with the determination of gait parameters using inertial measurement units (IMU). An IMU sensor incorporates three microelectromechanical sensors - triple-axis gyroscope, accelerometer and magnetometer. A standard experimental setup for the observation of the locomotion system using seven Xsens MTw sensors was developed. They are applied to the lower limbs and the pelvis of the subject. The synchronization of data from all sensor components (gyroscope, accelerometer and magnetometer) as well as the onboard estimation of the orientation is provided by the Xsens and Adwinda hard-and software. The strapped down data are received with a rate of 60 Hz. The output data of a single IMU sensor allow motion analysis of the sensor unit itself as well as the motion of the limb where the sensor is mounted to. Stable and reliable algorithms processing the gait data and calculating gait features of a single sensor are developed and evaluated. These algorithms are based on precise determination of each gait cycle. In the middle of stance phase the foot is not moving. It stands on the floor and, following, the initial conditions for the calculation of foot velocities and distances by integration are predetermined. Various features of the gait cycle as well as e.g. dependencies in between features or on the gait velocities are investigated. The application of seven sensors to the limbs of the locomotion system provides measurements of their 3D motion observed in an inertial coordinate system. The limbs are parts of skeleton and interconnected by joints. Introducing a skeleton model, the quality of measurements is evaluated and improved. Joint angles, symmetry ratios and other gait parameters are determined. These results can be used for analysis of the gait of any subject as well as of any cohort.