A constrained nonlinear optimization solution for 3D orientation estimation of the human limb

The aim of this research is to improve the methodology for human limb attitude estimation using wearable sensors. The orientations of these limb segments are measured using inertial/magnetic sensor modules. Such sensor modules typically contain a triad of orthogonally mounted accelerometer and magnetometer. The accelerometer is used to measure the gravity vector that is relative to the coordinated frame of the sensor module. Magnetometer serve a similar function for a local magnetic vector. Based on these two observations, we can formulate a constrained nonlinear orientation estimation of the human upper kinematics, based on the Wahba problem formulation. Several algorithms had been proposed earlier for solving this problem but all those currently solutions focused on minimizing the lost function (or cost function) numerically. As a result, these methods led to errors owing to unexpected noise, internal and external impacts such as, drift, vibration, force changes, etc. Hence, to mitigate this issue and enhance the effect of the solutions in tackling Wahba problem, a reachable workspace is defined in this work in addition to the current research algorithms.