Related Experiment Video
Updated: Jan 13, 2026
16:14
Trajectory Data Analyses for Pedestrian Space-time Activity Study
Published on: February 25, 2013
14.1K
An Improved Step Detection Algorithm for Indoor Navigation Problems with Pre-Determined Types of Activity.
Michał Zieliński1, Andrzej Chybicki2, Aleksandra Borsuk1
1Faculty of Electronics, Telecommunications and Informatics, Gdańsk University of Technology, 80-233 Gdańsk, Poland.
Sensors (Basel, Switzerland)
|October 29, 2025
Summary
This study presents a smartphone-based indoor navigation system using inertial sensors and a long short-term memory network for accurate step detection. Activity-specific models achieved up to 96% accuracy, improving pedestrian dead reckoning in complex environments.
Area of Science:
- Computer Science
- Robotics
- Signal Processing
Background:
- Global Positioning System (GPS) is unreliable indoors.
- Indoor navigation (IN) is crucial for public spaces like hospitals and airports.
- Pedestrian dead reckoning (PDR) relies on accurate step detection and counting.
Purpose of the Study:
- To develop a smartphone-based indoor positioning system.
- To leverage inertial sensor data for step detection and counting.
- To improve the accuracy of indoor navigation systems.
Main Methods:
- Utilized smartphone inertial sensor data.
- Employed a long short-term memory (LSTM) network for step pattern recognition.
- Trained generalized and scenario-specific models for different movement types.
Main Results:
- Achieved an average step detection accuracy of 93% with a generalized model.
- Scenario-specific models reached up to 96% accuracy for complex movements.
- Reduced false positives by specializing models for specific activities.
Conclusions:
- Smartphone-based PDR using LSTM networks is effective for indoor navigation.
- Activity-specific training enhances accuracy in complex indoor environments.
- Further optimization is needed for real-time mobile deployment latency.
