This Is AuburnElectronic Theses and Dissertations

A Loosely Coupled GNSS/PDR Integration Approach for Pedestrian Navigation





Jones, Connor

Type of Degree

Master's Thesis


Mechanical Engineering


Pedestrian navigation systems embody small, lightweight hardware, techniques, and biomechanical information to provide positioning information in an architecture that substitutes, or supplements, traditional systems such as GNSS. Two unique challenges of pedestrian navigation systems are the hardware size and weight constraints to keep the user comfortable, while the other challenge is the ability to provide accuracy and stability of the navigation solution in certain environments. Existing systems rely on the availability of small, lightweight, GNSS and inertial hardware for position, velocity, and attitude information in challenging environments. This thesis presents methods of integrating GNSS with a torso-mounted IMU to estimate three physical parameters of the user and the system hardware as a means of providing longer stability of the position, velocity, and attitude estimates when GNSS is no longer available. The presented methods will include showcasing a method of estimating the user's step length with existing models, as well as a new model; a method of hardware misalignment compensation for heading estimation; and an approach to detecting erroneous magnetometer measurements to reduce errors in the user's heading. GNSS is utilized in conjunction with the IMU to provide discrete step length pseudo-measurements for the user's step length estimation; GNSS course measurements will be used to estimate heading misalignment between the user and the IMU; and, finally, a threshold metric of the magnetometer measurements is used to compensate for errors in heading that would occur from perturbed magnetic field measurements. Performance analyses of each method is shown using real data with simulated GNSS outages. The methods are implemented using IMU data from a Vectornav 9-DoF VN-100 and a Ublox EVK-7 GNSS receiver for some data sets, and a Ublox ZED-F9P GNSS receiver for other data sets. Conclusions drawn from results of each method implementation are discussed and summarized.