Kinematic Model-Based Pedestrian Dead Reckoning for Heading Correction and Lower Body Motion Tracking.

Area of Science:

• Robotics
• Sensor Fusion
• Human Motion Analysis

Background:

• Pedestrian dead reckoning (PDR) using Zero Velocity Updates (ZUPT) offers short-term accuracy but suffers from heading instability due to magnetic disturbances.
• Inertial Measurement Units (IMUs) on the waist provide heading and position but require integration with other sensor data for improved reliability.

Purpose of the Study:

• To develop an enhanced method for pedestrian heading and position estimation.
• To improve the accuracy and stability of pedestrian dead reckoning (PDR) by addressing heading errors and cumulative position drift.

Main Methods:

• Fusing ZUPT-based PDR with lower human body kinematic constraints using an Extended Kalman Filter (EKF).
• Integrating foot-based ZUPT measurements with IMU data from the waist, utilizing a kinematic model of the lower body.
• Implementing an update-reupdate technique to enhance error state observability and correct position errors in foot-mounted sensors.

Main Results:

• Achieved a 1.25% Return Position Error (RPE) relative to walking distance.
• Demonstrated enhanced heading accuracy by mitigating magnetic disturbances.
• Improved overall position accuracy and stability over time compared to traditional ZUPT-based PDR.

Conclusions:

• The proposed fusion method significantly improves pedestrian heading and position estimation accuracy.
• The kinematic constraints of the lower human body provide valuable measurements for enhancing PDR systems.
• The technique offers a robust solution for motion tracking applications, providing comprehensive lower body kinematic information.