Landmark-Based Drift Compensation Algorithm for Inertial Pedestrian Navigation.

Area of Science:

• Robotics and Navigation
• Sensor Data Fusion
• Inertial Navigation Systems

Background:

• Pedestrian navigation heavily relies on inertial sensors like accelerometers and gyroscopes.
• Low-cost inertial sensors suffer from noise, leading to significant orientation estimation errors, especially in yaw angle (drift).
• Existing drift compensation methods often correct position but not the fundamental yaw angle estimation errors.

Purpose of the Study:

• To develop a novel, seamless, landmark-based drift compensation algorithm for pedestrian navigation using only inertial measurements.
• To address the limitations of current algorithms by directly correcting yaw angle estimation errors.
• To improve the accuracy and reliability of pedestrian trajectory estimation.

Main Methods:

• A landmark detection algorithm was developed to identify features like corners and stairs from inertial data.
• An association algorithm was created to link re-visited landmarks, considering positional uncertainty.
• The detected drift is computed from landmark associations and applied in post-processing to correct yaw angle estimation.

Main Results:

• The algorithm reliably detects landmarks (corners, stairs) using only inertial measurements, requiring no user input.
• Successful association of re-visited landmarks was achieved, accounting for position uncertainty.
• Post-processing using computed drift resulted in significantly reduced yaw angle errors and drift-compensated trajectories.
• The method demonstrated effectiveness with simulated biases and medium-cost gyroscopes in diverse 3D environments.

Conclusions:

• The proposed landmark-based algorithm effectively compensates for yaw angle drift in pedestrian navigation systems.
• By directly addressing yaw drift using inertial data and landmark associations, trajectory accuracy is substantially improved.
• This approach eliminates the need for external sensors or user interaction, offering a seamless and robust solution.