Heading Estimation for Indoor Pedestrian Navigation Using a Smartphone in the Pocket

Area of Science:

• Computer Science
• Robotics
• Navigation Systems

Background:

• Smartphone heading estimation is crucial for indoor pedestrian navigation.
• Challenges include dynamic device orientation and indoor magnetic distortions.
• Existing methods struggle with accuracy and feasibility in real-world scenarios.

Purpose of the Study:

• To develop a robust heading estimation approach for smartphones in pockets.
• To address challenges of changing device coordinate systems and magnetic perturbations.
• To improve accuracy and feasibility of indoor pedestrian navigation.

Main Methods:

• A novel approach using a rotation matrix to project acceleration signals into a reference coordinate system (RCS).
• Principal Component Analysis (PCA) applied to horizontal acceleration signals for local walking direction extraction.
• A calibration process without compass readings and a turn detection algorithm for global direction translation and accuracy enhancement.

Main Results:

• The proposed method accurately estimates the horizontal plane of acceleration signals in the RCS.
• Principal Component Analysis (PCA) effectively extracts local walking directions.
• The developed calibration and turn detection algorithms enhance global heading accuracy.
• Experimental results demonstrate superior accuracy and feasibility compared to uDirect and traditional PCA-based methods.

Conclusions:

• The novel rotation matrix and PCA-based heading estimation approach effectively overcomes key challenges in indoor smartphone navigation.
• The method provides a feasible and accurate solution for pedestrian navigation systems.
• This technique offers a significant advancement for reliable indoor localization.