Pedestrian Dead Reckoning Based on Frequency Self-Synchronization and Body Kinematics

A novel pedestrian dead reckoning method conceived to be used with sensors freely positioned not too far from the waist level is presented. Attitude and heading reference systems already built in in nowadays inertial measurements units (IMUs) are exploited to cast the sampled data into a global reference coordinate system, where human gait analysis can be used to figure out the motion related to each single step. In particular, vertical accelerations are processed by means of a phase locked loop to detect the pace and the steps, and then the step length is computed exploiting an empirical piecewise linear relationship with the pace, while the geometrical features of the planar acceleration are used to estimate the stride heading, based on the waist kinematics. Experiments show the good results of the proposed algorithm when using both a low-cost IMU embedded in a smartphone and a more expensive stand-alone device, highlighting the method robustness with respect to the implementing hardware.