Analysis of spatial parameters during gait with magneto-inertial sensors and infrared proximity sensors

The aim of this thesis is to estimate the base of support (BoS) during gait, defined as the area over which the body is supported during the double-support phase. To reach this goal, both absolute and relative orientation and position of the feet are required. The relative orientation is achieved by sensor fusion algorithms that combine information from accelerometers, gyroscopes and magnetometers. In fact, in this study the orientation is obtained by using Magneto-Inertial Measurement Units (MIMUs) that are suitable for measurements outside of motion analysis laboratory and over extended period of time. However, the use of inertial technology itself does not allow to obtain the absolute position estimation due to the lack of a global reference with fixed origin. These problems can be overcome by using additional distance sensors (DS). We plan to adopt infrared time-of-flight proximity distance sensors (IR ToF), which guarantee a high output data rates and allow that transmitter and receiver are embedded in the same sensor. First of all, to refer both feet to the same system, orientation and position with respect to an absolute coordinate system are required. This is the pre-requisite of further step width and BOS estimation. MIMUs provide the orientation relative to an absolute frame, which is identified by the direction of the gravity and the Earth magnetic field. Although, the position information cannot be achieved by MIMUs, which provide only the relative displacement with respect to the initial position of the sensors. In fact, the single foot orientation will be referred to the a global frame that is not fixed in a defined origin point. DSs overcome this limitation, offering the relative distance between feet and, thus, connecting both feet coordinate systems and enabling to provide absolute positions. To accurately estimate the 3D orientation, complementary properties of accelerometer, magnetometer and gyroscope are exploited and their data are combined in a sensor fusion algorithm. MIMU systems suffer from measurement errors and integration drifts, which can limit displacement and orientation assessment during long-term measurements. Thus, the study begins with choosing a suitable way to estimate the orientation through MIMUs and then compensating the errors. Consequently, the orientation of each foot with respect to the global non-fixed frame is obtained. To find the trajectory and displacement of the foot during time, the measured gravity-free acceleration is double time integrated and temporal parameters have to be calculated to impose the right initial and final conditions at every gait cycle. The zero velocity update is used to reduce the integration error, imposing that the velocity must be zero during the foot flat phase, when the foot is still on the ground. IR ToF sensors measure the time an electromagnetic wave needs to travel a distance. Once distances between the swing foot and the stance foot are collected, we can solve a scanning problem creating a model of the scanned foot and, then, describe the position of both feet with respect to the same previously fixed global reference frame. Thus, absolute positions of both feet are obtained. The tests for methodology validation will include experiments using mechanical analogues replicating the gait mechanism and on a healthy subject. The validation is made with stereophotogrammetry, which is considered the gold standard in gait analysis.