A Robust Adaptive Filtering Framework for Smartphone GNSS/PDR-Integrated Positioning

Area of Science:

• Navigation Systems
• Signal Processing
• Geomatics Engineering

Background:

• Outdoor pedestrian positioning via smartphones faces challenges in urban canyons due to degraded Global Navigation Satellite System (GNSS) signals and cumulative errors in Pedestrian Dead Reckoning (PDR).
• Existing methods struggle with accuracy and reliability in environments with poor satellite visibility.

Purpose of the Study:

• To develop a novel fusion method for accurate and continuous outdoor pedestrian positioning using smartphones.
• To enhance the robustness and accuracy of navigation in challenging urban environments.

Main Methods:

• A two-stage filtering architecture based on a Robust Adaptive Cubature Kalman Filter (RACKF) was proposed.
• The first stage uses a quaternion-based RACKF for sensor fusion (gyroscope, magnetometer) for heading estimation.
• The second stage fuses GNSS observations with an enhanced 3D PDR solution, incorporating adaptive noise estimation, outlier suppression (M-estimator), and differential barometric height.

Main Results:

• The proposed method achieved a horizontal positioning accuracy of 3.28 m (RMSE).
• It demonstrated superior performance compared to standalone GNSS.
• Significant improvements were observed in 3D PDR accuracy, by 25.97% and 10.39% respectively.

Conclusions:

• The novel RACKF-based fusion method provides a practical, infrastructure-free solution for robust smartphone-based outdoor navigation.
• The adaptive noise estimation and outlier suppression mechanisms enhance positioning reliability in complex environments.
• This approach effectively overcomes GNSS signal degradation and PDR cumulative errors.