Quaternion-based unscented Kalman filter for accurate indoor heading estimation using wearable multi-sensor system.

Area of Science:

• Robotics and Navigation
• Sensor Fusion
• Micro-electromechanical Systems (MEMS)

Background:

• Inertial navigation systems (INS) offer high reliability but face challenges in heading estimation.
• Magnetometers are susceptible to magnetic disturbances, while MEMS gyroscopes suffer from time-dependent inaccuracies.
• Accurate indoor heading estimation is crucial for various applications, including robotics and personal navigation.

Purpose of the Study:

• To design a compact and cost-effective wearable multi-sensor system for high-accuracy indoor heading estimation.
• To address the limitations of individual sensors (magnetometers and gyroscopes) in noisy indoor environments.
• To validate the system's performance on both pedestrian and unmanned aerial vehicle (UAV) platforms.

Main Methods:

• Development of a wearable multi-sensor system integrating a three-axis accelerometer, three single-axis gyroscopes, and a three-axis magnetometer.
• Implementation of a quaternion-based unscented Kalman filter (UKF) algorithm for sensor fusion and heading calculation.
• Experimental validation in a college building environment with the system worn by a pedestrian and mounted on a quadrotor UAV.

Main Results:

• The wearable multi-sensor system demonstrated effective indoor heading estimation.
• Mean heading estimation errors were below 10° for the pedestrian platform.
• Mean heading estimation errors were below 5° for the quadrotor UAV platform, compared to the reference path.

Conclusions:

• The proposed quaternion-based UKF approach effectively fuses data from multiple sensors for robust indoor heading estimation.
• The compact and low-cost wearable system provides a viable solution for accurate heading determination in challenging indoor environments.
• The system's performance on both pedestrian and UAV platforms highlights its versatility and potential for widespread application.