Bounding inertial drift with human gait dynamics for personal navigation

Abstract

The integration of inertial measurements to estimate position and orientation is limited in its accuracy by the biases in the inertial instruments. The gyroscopes and accelerometers used in personal navigation systems do not have extremely high fidelity. The drift in position estimates grows as the cube of time elapsed without absolute measurements such as GPS to bound it. However, one source of inertial aiding has largely been overlooked, which is the unique dynamics of gait for each individual. Based on the record of inertial measurements in walking or running, the relationships between stride length and stride interval for individuals is developed. We designed an integrated system with corrected human gait model to limit the drift of inertial integration with the assistance of this relationship. Analysing in individual step and corrected with gait model can effectively minimize the error of position estimating.