Optimizing Pedestrian Dead Reckoning (PDR) Accuracy through Motion Behavior Recognition and Dynamic Step Length Adjustment

Abstract

The Pedestrian Dead Reckoning (PDR) algorithm is a method for determining pedestrian position by integrating inertial data from an inertial measurement unit (IMU), the subject's movement behavior, and environmental information. The accuracy of the PDR algorithm relies on key system parameters, such as step count, step length, and the subject's movement behavior. This study employs combined acceleration data to count pedestrian steps, utilizes the K-nearest neighbor (KNN) method to estimate the subject's movement behavior, and dynamically adjusts the step length estimation model based on the recognized behavior patterns. Consequently, this approach achieves high-precision pedestrian position estimation, maintaining an average error within 3% of the total distance traveled.