Model Based Inertial Sensing for Measuring the Kinematics of Sit-to- Stand Motion

In this paper a method for measuring kinematics of sit-to-stand movement using inertial sensors and human body model is presented. The proposed approach fuses data from inertial sensors (accelerometers and gyroscopes) and data from three-segment human body model using Extended Kalman filtering technique and in this way reduces inaccuracies associated with inertial sensors. Dynamic human body model is constructed based on principles of Lagrangian dynamics and incorporates shank, thigh and HAT (Head-ArmsTrunk) segments. The model equations, which are non-linear and coupled differential equations, are incorporated in the EKF to produce better kinematic parameter estimation of sit-to-stand motion. The moments needed in model equations (ankle, knee and hip moments) are calculated based on EKF last best estimate and Newton-Euler inverse dynamic approach. Outputs from EKF are segmental angles, angular rates of change and angular accelerations. These parameters fully describe the motion kinematics of human body. The performance of the method is verified by reference measurements acquired with optical motion measurement system Optotrak. Obtained results are presented and discussed. Conclusions are drawn and guidelines for future method improvement are suggested. Key-Words: inertial sensors, human dynamic body model, extended kalman filter, sit-to-stand transfer