Comparison of Computational Efficiency of Magneto Inertial Sensor Fusion Algorithms for ChakaMo

During human motion tracking there are several tools that allow the analysis of joints during motor tasks. One of these devices is the magneto-inertial sensor that estimates the orientation of each limb segment and consequently the articulation kinematics. The orientation estimation with the magneto-inertial sensors relies heavily on the combined calculation with data from accelerometer, gyroscope and magnetometer, for which sensor fusion algorithms are used. At the moment there is no generic algorithm that works for all motor tasks, therefore, we compared in this work four sensor fusion algorithms to find out which one has the best behavior during a knee flexion and extension phantom simulation. After a complex comparison based on the computational efficiency, implementation easiness and minor root mean square error, the Guo algorithm is chosen over Madgwick and two Valenti options. The selected algorithm is used in the development of ChakaMo, a clinical tool for 3D analysis of the knee during step up and down motor task.