Research on the Error Correction Method of Strap-down Inertial Navigation System for Pedestrian Positioning and Navigation

Global Navigation Satellite Systems (GNSS) experience drift and inability to provide accurate positioning services when obstructed and subjected to electromagnetic interference. Hence, Inertial Navigation Systems (INS) are widely utilized in indoor and heavily obstructed outdoor scenarios, such as fire rescue and railway personnel location, due to their advantages of lightweight, low power consumption, low cost, and interference resistance. However, the low frequency inertial measurement unit (IMU) suffers from the problem of low sampling data frequency and error dispersion over time, which will affect the final positioning accuracy of INS. To address these issues, this paper proposed an error correction method of strap-down Inertial Navigation System for pedestrian positioning and navigation, combining Piecewise Cubic Hermite Interpolation Polynomial (PCHIP), multi-threshold zero velocity detection and Cubature Kalman Filter (CKF). The experimental results demonstrate that this method can effectively correct the cumulative error of Inertial Navigation System. During the test with a pedestrian walking time of 85 seconds, the cumulative error was reduced to 0.1703 meters, limiting the cumulative error to 0.28% of the entire walking distance.

I. Introduction

Inertial Navigation Systems (INS) are autonomous positioning systems that do not rely on external information or emit energy to the outside, often used in scenarios where Global Navigation Satellite Systems (GNSS) fail, such as indoor environments or in the presence of electromagnetic interference [1][2]. INS devices are primarily categorized into platform-based INS and strap-down INS. Platform-based INS devices are large in size and high in cost, making them unsuitable for personnel positioning. Strap-down INS devices, on the other hand, are compact and portable, mostly composed of Micro Electro Mechanical Systems (MEMS) inertial sensors such as accelerometers, gyroscopes, and magnetometers, also known as Inertial Measurement Units (IMU). Low-frequency IMUs have low sampling data frequency and large errors, leading to exponential error divergence over time, thereby greatly increasing the difficulty of subsequent data processing [3].