Pedestrian Navigation System with Trinal-IMUs for Drastic Motions

Area of Science:

• Biomechanics
• Robotics
• Sensor Fusion
• Navigation Systems

Background:

• Global Positioning System (GPS) is unreliable in indoor or obstructed environments.
• Inertial sensor-based pedestrian navigation faces challenges in step length estimation and heading drift, especially during dynamic movements like running or sprinting.
• Accurate pedestrian positioning is critical for applications such as fire rescue and indoor security.

Purpose of the Study:

• To propose a trinal-node inertial measurement unit (IMU)-based system for simultaneous localization and occupation grid mapping.
• To enhance the accuracy of step length estimation during various gaits, including walking, running, and sprinting.
• To improve pedestrian pose estimation and overall localization accuracy in GPS-denied environments.

Main Methods:

• A trinal-node system utilizing two thigh-worn IMUs and one waist-worn IMU.
• Gait detection and segmentation via zero-crossing detection of the difference in thigh pitch angle.
• A piecewise function correlating step length with the probability distribution of waist horizontal acceleration for accurate estimation.
• Simultaneous localization and mapping (SLAM) based on occupancy grids, incorporating historical trajectory data for pose refinement.

Main Results:

• The proposed system accurately identifies and segments gait cycles across walking, running, and sprinting motions.
• Average step length estimation error was below 3.58% of the total travel distance for speeds ranging from 1.23 m/s to 3.92 m/s.
• Localization error was less than 5 meters within a 2643.2 m² single-story building when combined with the occupancy grid SLAM method.

Conclusions:

• The trinal-node inertial navigation system offers a promising solution for accurate pedestrian positioning in GPS-unavailable environments.
• The developed methods significantly improve step length estimation and reduce localization errors, even during high-speed and dynamic movements.
• This approach enhances the feasibility of inertial sensor-based navigation for critical applications like rescue and security operations.