Wearable Human Upper Limb Motion Tracking System: A Kinematic-Based Filtering Approach

This work aims to define a method for estimating the position and the orientation of the human limbs. Accurately tracking the movements of the human body is a crucial operation for what concerns several applications, like, for example, the remote motor rehabilitation or the creation of a persons' virtual model in a video game, to mention just a few. In particular, an inertial measurement unit (IMU) sensor network is exploited to measure the acceleration and the angular velocity of the upper limb parts, which are combined, along with a kinematic model of the arm, in a sensor fusion algorithm to estimate the orientation of the limb up to the wrist. For the purpose, a 7 degrees-of-mobility kinematic chain is used to model a limb, the three links of which represent the clavicle, the upper arm and the forearm. Being the origin of the global reference frame placed in the thorax, the sternoclavicular joint, the shoulder and the elbow are modeled as multiple rotational and spherical joints. The orientation of the links in space is represented by quaternions in order to overcome the matrix inversion issues proper of minimal attitude representations. The pose of the links is then estimated by means of an Extended Kalman Filter (EKF), in which the accelerations of a forward kinematic model are compared with the data provided by the sensors.