Inertial Indoor Pedestrian Navigation Based on Cascade Filtering Integrated INS/Map Information.

Area of Science:

• Robotics and Navigation
• Sensor Fusion
• Geomatics Engineering

Background:

• Indoor pedestrian positioning is crucial for applications like fire rescue and navigation.
• Inertial Measurement Unit (IMU)-based systems offer cost-effective, equipment-free positioning.
• IMU systems suffer from cumulative errors, degrading positioning accuracy over time.

Purpose of the Study:

• To develop a novel cascade filtering algorithm for accurate indoor pedestrian positioning.
• To correct accumulated errors in IMU-based positioning using limited map information.
• To improve both 2D and 3D positioning accuracy, including altitude estimation.

Main Methods:

• A lower filter employing zero-velocity updates and Extended Complementary Filtering (ECF) for initial trajectory estimation.
• An upper filter utilizing a Particle Filter (PF) integrated with map data to refine heading and stride length.
• A clustering-based floor discrimination method to address barometer instability for altitude correction.

Main Results:

• Achieved a Root Mean Square Error (RMSE) of 1.35 m in 2D positioning.
• Demonstrated a low end-point error of 2.45 m over a 536.5 m trajectory in 3D positioning, including stair climbing.
• Successfully mitigated altitude estimation errors caused by environmental pressure and temperature fluctuations.

Conclusions:

• The proposed cascade filtering algorithm significantly enhances the accuracy of IMU-based indoor pedestrian positioning.
• The integration of map information and advanced filtering techniques effectively corrects cumulative errors.
• The method provides a robust and accurate solution for 3D indoor navigation, even in complex environments.