Robust Indoor Pedestrian Backtracking Using Magnetic Signatures and Inertial Data

Area of Science:

• Assistive Technology
• Human-Computer Interaction
• Robotics

Background:

• Visually impaired individuals face significant challenges navigating unfamiliar indoor environments.
• Current navigation aids often rely on pre-existing maps or external infrastructure, limiting their applicability.
• Effective wayfinding, especially backtracking, is crucial for independent mobility.

Purpose of the Study:

• To develop a hands-free indoor navigation system for visually impaired individuals.
• To enable reliable backtracking of previously traversed paths without external infrastructure.
• To enhance positioning accuracy using a hybrid machine learning approach.

Main Methods:

• A smartphone-based system was developed for indoor navigation.
• A hybrid matching method integrating machine learning was employed for enhanced positioning.
• The system was designed to operate without pre-existing maps or external infrastructure.
• Testing involved datasets from visually impaired individuals.

Main Results:

• The proposed system demonstrated potential for reliable backtracking assistance.
• Machine learning integration improved positioning accuracy, addressing real-world challenges.
• The hands-free approach proved effective for visually impaired users.

Conclusions:

• The developed navigation solution offers a promising approach for visually impaired individuals.
• The system's ability to facilitate path backtracking enhances user independence.
• Future work could further refine the machine learning model for even greater accuracy.